Episode 5: Cloudera – Machine Learning & Big Data with Mike Olson

Mike Olson is the Co-founder and CSO of Cloudera, the big data analytics platform that delivers Enterprise tools leveraging Apache Hadoop. In this episode, Mike describes how Cloudera contributes to the open source community while also holding back enough proprietary IP to build one of the most successful open source software businesses of all time.



Michael Schwartz: Welcome to Open Source Underdogs, the podcast where we dig into the business models of the best open source software companies in the world.

Today I’m excited to be with Mike Olson, Founder and Chief Strategy Officer at Cloudera.

Cloudera provides Enterprise tools for data engineering, data warehousing, machine learning, and analytics that leverage the open source Apache Hadoop platform.

Mike Olson, thank you so much for joining the podcast today.

Mike Olson: Michael thanks for coming by, I’m excited to talk.

How Did Cloudera Get Started?

Michael Schwartz: How did Cloudera get started?

Mike Olson: The company’s ten years and a few months old at this point. We started in 2008, early summer.

The conviction that I had, along with my three co-founders Amr Awadallah, Christophe Bisciglia, and Jeff Hammerbacher. The conviction we all had was that big data was going to be a big deal.

So if you looked at Yahoo, and Google, and Facebook and others, they were collecting and analyzing data at enormous scale – much bigger than banks or hospitals were doing at the time. And they had invented a collection of new tools to do that.

So Apache Hadoop was sort of the foundational project of this ecosystem. We all believed that banks, hospitals, insurance companies would want to get lots of data and then analyze it in powerful new ways. And this wonderful open source Hadoop project was ideally designed for that.

So our idea was let’s bring those capabilities to traditional Enterprises.

Now I had had a long career in database technology, I had worked for Informix, and Oracle, and a bunch of other database startups you never heard of. I had a long career in open source, so I had worked at Berkeley Unix, on Postgres, Berkeley DB.

So I understood the big Enterprise data consumption behavior, and I kind of got open source. I didn’t know anything about Hadoop.

I had no code running in there, I had not developed it. But I understood what it did, how it worked. And my co-founders had all been actively involved in its development and its use at Facebook, Yahoo, Google.

So together I think we were uniquely positioned at that time to start the business and to try to bring this web technology to traditional Enterprises.

Cloudera Customers

Michael Schwartz: Who’s using Cloudera’s product today?

Mike Olson: Let me talk about what we’ve built for the market, because it’s been heavily influenced by our target customer; and that’ll give me a chance to explain then, who we chose, why and so on.

When we started the company, all we had was Apache Hadoop. That is two open source components, a file system DTFS, and distributed processing engine, MapReduce.

Those two things together were what Google invented for big data processing and what took over the consumer internet. Fast forward ten years, we have 26 different open source projects in the bundle now.

In general all of the new stuff has swiped the fundamental design ideas of Hadoop. Scale out, so deploy and run on a whole bunch of servers, and bring the processing to the data. So MapReduce let you plow through huge amounts of data by every server basically running out tiny little plow-through job on its own, and then you combine the results at the end.

Well, there are other processing engines available now, there is Apache Impala for high-performance distributed SQL analysis. There’s Apache Spark for stream processing, and for marshalling like model training in machine learning. There’s the SlrCloud Lucene search engine that does document search and so on.

So, a whole bunch of scale-out projects now can collaborate on the data on that cluster, on those servers. So you don’t need to make multiple copies of the day, you can do lots of different processing at massive scale by taking advantage of all of the CPUs that you have attached to the storage.

So, platform’s gotten a lot more interesting and handles way more analytic work less than ever before. We’ve built a collection of services that compliment that core open source offering.

Data governance, compliance, living up to regulatory regimes in these big industries like healthcare and financial services, we built tools on the platform that let administrators, and chief information officers, and the privacy team be sure that the rules are getting followed and policies were enforced.

We’ve got high performance security and encryption services, deploy, and manage this infrastructure at massive scale – if you’ve got a thousand computers running a mission-critical job how do you ensure that the jobs going to finish on time, and so on.

So we’ve built a collection of large Enterprise-focused software compliments to that open source platform. The data storage, the processing, and so on, all of that is in the open source and were substantial contributors to that ecosystem, and we benefit from the great workout that ecosystem.

And then we build a product that adds some proprietary IP. That is not, that is differentiated and not nearly as easy for a competitor to pick up and use against us, as the open source would otherwise be. So we maintain some differentiated IP and some reasons for customers to buy from us.

We chose those capabilities: Regulatory support, operations at massive scale, compliance, and so on, security, because we aim to sell to large Enterprises.

Banks and hospitals and insurance companies have lots of data, complex businesses with lots of opportunities to do analysis, so they’re good customers, and crassly they got a lot of money.

So we designed our product strategy, and we did a lot of really innovative stuff expressly to go after, kind of that same large Enterprise buyer that I had been servicing for my entire career as a relational database industry player.


Michael Schwartz: There must be a lot of competition in this market. How do you differentiate Cloudera from some of the other offerings that are out there?

Mike Olson: In 2008 when we started the company nobody had heard of Hadoop.

It was just, you know, it was some nerd technology, if you were an engineer at Yahoo you were up to date. But the industry at large hadn’t heard of it.

The meme of Big Data didn’t even exist yet, so those words weren’t being used to apply to any kind of business problem.

That changed in a hurry. So many competitors have entered the market, lots of folks have recognized the value of Big Data. A number of companies have even stepped in and joined us in delivering solutions that build on the Apache Hadoop ecosystem.

Look, I think there’s a bunch of reasons we’re good, we’re deeply involved in the open source community in order to take advantage of and help drive that benefit, we’ve got to do that. But our differentiation, how we set ourselves apart, is really based on how we been thinking about the customer.

So large Enterprises need the security governance compliance and regulatory support we’ve got. We genuinely believe that, given the decade that we’ve been building the product and the expertise we got, we’re differentiated there.

In addition however, our large Enterprise clients have data centers, and we think they’re going to have data center for a long time, so they want to be able to run stuff on premises.

They need to be able to take advantage of the public cloud, so spin jobs up on Google or on Amazon or on Microsoft Azure. And then, they’d like to be able to move those workloads around among those places – on prem, the various public cloud providers, as business requirements demand.

We support support this sort of, hybrid on-prem and cloud, and also multi-cloud, you can move among the cloud vendors capability, so that our customers get all that security, governance, compliance, regulatory support, and aren’t locked into a single cloud provider.

There are outstanding offerings that are single cloud only, so you know Amazon’s got a rich collection of great data services, and we integrate with and complement those.

But the promise we make our CIO customers is, you can deploy with confidence on any of those venues because if you decide later to move, your applications don’t change, you don’t, you haven’t coded to any sort of built-in proprietary, native cloud API’s. You’re building on the open source ecosystem of Hadoop, and your app will move easily.

Customer Interaction

Michael Schwartz: Are there any grades of interaction, maybe for smaller customers or the community?

Mike Olson: So of the 26 open source projects that we distribute, some were ingested from the open source community, created outside of Cloudera entirely – so Apache Spark’s a great example. Now we’re major contributors in the Spark ecosystem as well, but that was created at Berkeley and then released through the Apache Software Foundation.

One of the two original founders of the Hadoop project, Doug Cutting, actually works at Cloudera so you know, that predated the company, but obviously we were engaged in that from the very earliest days.

And we’ve created open source projects. So a great example is Apache Impala, our distributed query processing engine. Or Apache Kudu, an IoT-based storage engine that were conceived and originally developed here, and released through the Apache Software Foundation to the open source community, and we’ve built communities around that.

We benefit from the great work of the global open source development community. We earn our credibility by contributing to that community as well, so both of those are important.

We don’t really engage commercially with say, small companies with small data problems. Remember, we’re aimed at these large organizations and regulated industries with complex business.

So we really sell to very, very large Enterprises and we focus on those problems, and our interactions with those customers is aimed at basically that kind of buyer.

So we don’t really do much with open source consumption of the platform, although a great deal of it goes on.

We’re concentrating on monetizing and easing the adoption of the open source tech by those big guys.

What Goes In Open Source?

Michael Schwartz: How do you decide how much to invest in the open source projects versus your commercial project?

Mike Olson: This is going to sound glib, actually a whole bunch of thought has gone into it. But very briefly if it’s about data storage, data analysis, data processing – CIO’s don’t want to be locked in.

They don’t want a single vendor proprietary solution.

I mean, I spent a bunch of my career working for great big database vendors and we taught CIO’s that single vendor proprietary lock-in was a mistake. So everybody’s learned, they don’t want me to be able to turn off their data access, they don’t want me to be able to shut down their analytic workload.

So if it’s data storage, data process, data analysis really it needs to be in the open source.

If it’s addressing the unique requirements of a large Enterprise, like if you need to be able to answer the demands of regulators, or if law enforcement shows up and you’ve got to do a data lineage work.

Well, look man, that’s fair game. Because first of all the open source community isn’t going to spend a lot of time on that kind of problem. And 2nd it’s expensive and difficult to develop those solutions, and so I should legitimately be able to be paid for those.

And so in general we’ve got a philosophy on needs to be open source and what can be closed source, proprietary IP.


Michael Schwartz: Have you developed channel partners, or do you rely on the open source to be a distribution Channel?

Mike Olson: We’ve got good channel relationships with some of the big systems vendors.

So you know Dell is a fantastic partner, HP’s a fantastic partner, even Oracle and Teradata offer appliances that bundle the Cloudera platform and they’ll sell through their sales force to their customer base our IP on those appliances.

The global systems integrators are likewise a really good channel for us they turn up a bunch of opportunity.

The bulk of our revenue is from direct sales.

So we’ve got a substantial global sales force with concentrations in AMEA, in Asia-Pacific, and then the Americas and leaders identified there. And we have a bunch of direct sellers and technical folks in the field that engage with our large Enterprise customers.

Initial Sales Strategy

Michael Schwartz: When you were first getting started was it really hard to figure out how to manage the long sales cycle of a large Enterprise? How did it go in the early days?

Mike Olson: Bear in mind that had been the market that I sold two for my entire career, so I knew what to expect in terms of sale cycle. And I was an engineer to begin with right, but in middle-late 90s I stopped being an engineer and became or more and more field-focused.

Anyway, I was used to the buying cycle of that community. I’ll say that we got lucky in a weird way.

So you remember 2008 was the Global Financial Market meltdown. It had a really important knock-on effect, it was everybody was really afraid of what was going to happen further in the economy.

So large Enterprises were looking for ways to shed expense, and if they had to do innovative stuff to do it in novel and much less expensive ways.

So you know, it used to be that you could legitimately go charge somebody $40,000 to manage a terabyte of data. You know there’s a lot of terabytes of data in the world these days, and at 40,000 bucks a terabyte, there’s a huge penalty for having data.

Well this platform was designed to make lots of data cheap, to a cheap to accumulate, and then operate on. So the market crashed, a bunch of CIOs got very cost-conscious, and here we were with this open source foundation and a new way of building scale-out, distributed systems that was vastly cheaper than what came before.

Look I mean, the market crash was a disaster for a bunch of reasons but it was a little bit lucky for us. We tried to stay rigorously focused on what we thought customers needed and that varies by time.

You know the needs of customers today are very different from what they were a decade ago, because everybody gets open source now, right. Everybody gets big data.

We’ve evolved our product strategy along with the maturing market. And I’d, maybe it’s vain, I’d claim we helped to mature the market as we went.

Red Hat IoT Partnership

Michael Schwartz: One of the new trends is IoT. To expand a little bit on the partnership angle, we were reading about a partnership with Red Hat.

Do you see that as the new direction, that better solutions are needed?

Mike Olson: I actually do and unsurprisingly that’s why we got involved in, that’s why I’m so excited about the work we’re doing with Red Hat.

I’d take a step back, IoT as a broad secular trend, is a huge boon to those of us in the big data industry. Right, I mean that data got to come from somewhere.

It’s nice if we just put sensors all over the place, if we were able to ingest you know, stock trader market data at very fine grain, and then we can store huge amounts of it. So the more things there are on the internet making data, the better off that the big data platform is going to be. So, very excited about it on that basis.

If you know much about the Hadoop architecture, it’s kind of this big back-end, where data lands and then you analyze it. What we hadn’t done, what we don’t build, is the end-to-end architecture.

So how do you rely on to get data off of the sensor, or off of the device? What happens to it as it runs the network? Is their analysis and aggregation of data in flight?

Well it turns out the folks at Red Hat and a number of other industry partners, were thinking about those capabilities and taking advantage of their infrastructure software, to deliver some interesting services there.

If we can make our analytic platform a good destination for that data, and then train models using machine learning techniques to spot anomalous activity in those sensors in real-time – well that’s awesome, right?

So the combination of these technologies, really the partnership around IoT data, enables applications that we couldn’t have enabled on our own.

And then obviously, you know Red Hat’s been a fantastic partner in Enterprise software for for open source, forever and ever.

Other Partnerships

Michael Schwartz: Are there other partners that you think have been helpful?

Mike Olson: So in the IoT space in particular, Eurotech and the Eurotech Everyware family, is an important part of that partnership. Looking more broadly than just at that IoT activity though, like any platform software company we rely on a rich ecosystem of other companies in order to succeed.

So you squint your eyes, you’re allowed to think of a Cloudera Enterprise as a database. It’s just not a 1980s database it’s a 2018 database; lots more data with lots more powerful analytic capabilities. But you need the application that runs on that database.

I want to predict which customers will churn out of my mobile service. I want to spot fraudulent transactions in data flows. I want to guess which patient at my hospital is likeliest to be readmitted based on past behavior.

We don’t build that app but we have a collection of systems integrator partners, and independent software vendors that do build those applications. And they’re absolutely essential to our success, right – nobody buys the platform, everybody buys the application, but the platform has to be there to build the application on.

License Strategy

Michael Schwartz: One of the popular licensing strategies in open source is open core. Do you have some thoughts on that strategy, or maybe licensing in general?

Mike Olson: Couple things. So first of all – this is just a super fraught emotional area. People have very strongly held views. I’m pretty pragmatic. I deeply believe in the strategic value of open source.

You get this global community of really smart folks, anyone who cares enough about a problem can join the community and concentrate on solving that problem; and because they care a lot about it, they’re probably going to be better at it then someone you choose randomly.

So open source innovates in ways that proprietary software can’t because you can harness the whole planet, and you get to take advantage of distributed expertise. I love it.

The challenges of purely open source products is able to be picked up, and given away by any vendor who wants to do that.

So you can think of a mega vendor grabbing a collection of open source IP, pricing it at zero and then monetizing that by selling proprietary database that connects to it, or a whole bunch of expensive services or… if you invest a lot an open source development you run the risk of being commoditized when someone else takes that IP. And drives its cost way down, so you wind up competing with them.

Our decision to complement our open source platform with proprietary IP is not intended to lock customers in, its intended to lock competitors out.

I want to have some reason that customers will come to talk to me and not go talk to one of those big bad commoditizing vendors. And that’s why we build our product, that’s why we have the IP strategy that we do.

We’ll always have proprietary IP that complements the open core, the open source.

What that is will evolve over time as our business involves. And the open source ecosystem will also evolve over time.

People are going to innovate a new ways, maybe some stuff that used to only be able to get one way you’ll be able to get four or five different ways. But I believe it’s a good sustainable long-term model. Certainly it’s served as well in the decade since we started the business.


Michael Schwartz: Would it be oversimplifying to say that the strategy is to build proprietary tools around the open source?

Mike Olson: It’s a fair view. I mean if, if I use those terms in front of my marketing guys, they’d say but we build way more than tools, we build a platform. Fair enough, right.

But complementary code aimed at the specific requirements of large Enterprises, that compliments all the power, flexibility, scale of the open source ecosystem.

Closing Advice

Michael Schwartz: So if you are an entrepreneur who wanted to use open source as part of your business?

Mike Olson: I’d say first of all, you know, 20 years ago… Oh heck, man. Yeah 20 years ago, 30 years ago, when I was working on Postgres, when I was working on Berkeley DB, there was this question you know, what is the open source business model?

I think that question is poorly formed. A business model is actually complex construct. Open source is a really important component of strategic thinking.

It’s a great distributed development model, it’s a genius low-cost distribution model – anybody can download your software at very low cost on very low friction.

And those have a bunch of advantages, right. You need to think about how you’re going to get paid. So what is it that people will give you money for, and it can’t just be because you’re good at what you do, because sooner or later somebody else is going to get good at that too, and competition is going to be tough to maintain attractive margins.

So you need to be thoughtful about the unique value that you’re adding. You need to think about who the customer is, and what they require, and why they will uniquely buy from you.

The particulars of the license that you choose matter a lot.

The GPL is coercive in a way that a lot of people like but that freaks out, for example, a lot of cloud vendors. They won’t pick up GPL’d code because they’re concerned about getting infected with the GPL requirements on code that they developed. The Apache license I think is actually a great license and has some good IP protection around patents, but is much more permissive.

So you want to think about what the license requires relative to what you’re going to deliver to your customers, and why they’re going to buy, and how that’s going to stay defensible. I think there’s a lot of really good thinking on this.

Clearly Red Hat is world-class at building an open source business, but it’s much more than – what should I do with open source? You want to think about who you’re going to sell to, why they’re going to buy, and why you’ll be able to preserve differentiation, and your advantage long-term in that market.

Michael Schwartz: Wise words from Mike Olson. Thank you so much for sharing your thoughts with us.

Mike Olson: Thank you Michael, I really enjoyed it. Thank you for coming by.

Michael Schwartz: That’s it for episode five. Transcription and episode audio can be found on opensourceunderdogs.com.

Special thanks to the Linux Journal for co-sponsoring this podcast to the All Things Open conference for helping us publicize the launch.

Music from Broke for Free, by Chris Zabriskie and Lee Rosevere.

Production assistance from Natalie Lowe. Operational support from William Lowe. Thanks for the staff of Cloudera for schedule juggling.

Next week we’re leaving the Bay Area and we’re off to the Big Apple, where we’ll visit with Eliot Horowitz, one of the founders of MongoDB.

Thanks for tuning in.

Episode 4: Mattermost – Private Cloud Messaging Platform with Ian Tien

Ian Tien is Co-founder and CEO of Mattermost, an open source, private cloud alternative to proprietary SaaS messaging solutions. Mattermost is used for secure team communication at organizations like Intel, Samsung, and the United States Department of Energy. In this episode, Ian describes their Open Core business model, which seeks to monetize enterprise customers who need advanced deployment and security features.

Mattermost Transcript


Mike Schwartz: Welcome to Open Source Underdogs, the podcast where we interview leaders from successful open source software companies and bring their stories to you.

Today my guest is Ian Tien, Founder and CEO of Mattermost.

Mattermost is a highly-scalable chat platform for privacy-conscious organizations that want out of cloud-hosted platforms like Slack or Skype.

Ian, thanks for joining us today.

Ian Tien: Thanks for having me.

Mike Schwartz: Ian, why don’t you just start a little by telling your listeners your background and a little bit of the founding story of Mattermost.

Ian Tien: Sure. So I actually began my career at Microsoft, up in Redmond.

And I was working on the Microsoft Office Team and Engineering with my friend, Corey Hulen. And we worked on number SharePoint products that had business intelligence built-in, dashboarding, data warehousing, SQL Server. And, you know, had to really good time doing that.

I moved over to the product management side with Outlook.com and OneDrive. Had a pretty good time there. So you know this is early days, and from there I ended up going to to business school for a little bit.

And while in business school I started a video game company. So I was doing online games, it was a lot of fun. And I ran that for a few years after business school.

And what happened is while we were, you know running this very sort of fun game company, we were using a messaging service online, and it was it was running our entire company: All of our data was in it, all our analytics, all the conversations, all our analysis, our designs.

And one day, this platform that were using got bought by a large company. It was a small company, got bought by large company. And the quality of the application started going down.

It would crash it, would lose data, and you know, we were just really frustrated by it, and we tried to export. We tried to export and they wouldn’t let us leave; we couldn’t get our data out.

And when we stop paying the subscription, they would paywall us from our own information.

When this happen, there were people are team saying like, this is, this is sort of not okay. This is not the way that the world should work. Where you’ve got these SaaS services that

can really keep you all contained.

So what we did is we had over 10 million hours of messaging done in our video games. So we took some of the software that we use for messaging, we rewrote it little bit, and, you know we’re engineers, so we rewrote it a couple times.

And pretty soon we were using our own software to do collaboration and to sort of replace their the platform we were using. It was pretty basic.

But what happened is, in June 2015 we decided open source it. And it just really took off. We end up winning some open source awards, and it became pretty clear that this was a business.

So we pivoted the company, we got rid of the games, and we said, you know we’re going to build this new world of open source messaging, and that was the Mattermost open source project.

By March of 2016 we’d figure it out a way to sell a Enterprise Edition of the product on top of the open source platform. And it’s really worked.

So now we’re all over the world and we have a lot of customers. Two of the top three aerospace companies, two of the top three mobile phone companies, two of the top three US federal agencies. These are all our customers, and what happens is, they use our open source Mattermost Team Edition, which is available MIT-licensed, really easy to install, single Linux binary, works with MySQL or PostgreS[QL].

People get going right away, they see immediate value. They’re messaging each other. Their collaborating, they’re adding lots of integrations, all you need to curl command and a webhook, and you can build these sort of like mini systems.

So you got immediate value, whether it’s devops teams, or information security, or IT teams, or broader users. They really enjoy the product and prove that there’s value.

And as the deployment grows, they decide, you know what, we shouldn’t as a team host this ourselves, we should have central IT do it. And when they give it to Central IT, Central IT says well you want these compliance features: We want high availability, we want data retention, we want compliance, can you plug into our [backend], our global relay, and can you do e-discovery.

And the answer is all yes – but those features are in our commercial version. So that’s when the procurement cycle starts.

After they’ve proven the value, we’ve given it all away for free. When it’s time to scale, that’s where you have the sales discussion.

So that seems to be a really nice model and that’s why we’ve grown so fast in just a couple years.

Customer Segments

Mike Schwartz: Do you segment the customers all, in terms of marketing? Or are you really just waiting for customers to find the open source and then contact you?

Ian Tien: Our growth so far is really around the open source version being used, and adopted, and expanded. And there’s going to be customers sort of all across the board.

You know, our focus is largely on Enterprise, because can we sell high-availability and retention, it’s really those large organizations that find the most value out of it.

Then there’s also more security-conscious, sort of smaller companies and they’ll buy some the Enterprise features for one or two things that they like. But it’s really, we’re really focused on the Enterprise segment.

Control Of Data / Privacy?

Mike Schwartz: And do you find maybe that some of the Enterprise customers don’t feel comfortable putting their data on a cloud server especially, you know, confidential conversations between people?

Ian Tien: Our customers are privacy-conscious Enterprises, who are looking for a team collaboration solution that can increase the productivity while meeting their custom security and sovereignty needs.

So large Enterprises, financial services, in public sector, in government, in manufacturing. In what we think of as high-trust organizations, they’re really looking for solutions that they can control a hundred percent.

They know that there’s no third party monitoring, they know that all the custom security and compliance infrastructure that they bought they can integrate with our collaboration solution. And they know that, you know, they don’t have any external dependencies, right. If this has to be available all the time, they have total control.

And I would say a lot of customers actually are going to public cloud. They’re on Azure, they’re on AWS. A lot of the public sector are on the govcloud versions of those. Customers are very comfortable with cloud and Kubernetes, and Docker, and all these sort of cloud-friendly technologies.

What they prefer is that they control the system and not a vendor. So when it comes to privacy and control of the data, they have it a hundred percent.

Customer Interactions?

Mike Schwartz: How do you work with customers? What’s your experience was supporting customers? Do you work a lot one-on-one with customers? Or do you rely more on automated online systems to work with customers?

Ian Tien: The business model we’ve chosen is open core. So there’s going to be an open source version that’s free, and everyone has it, and they can enjoy it, MIT-licensed, really easy to use. And then there’s a commercial version that’s an add-on that gives you sort of Enterprise features.

So what we don’t do is we we don’t try to monetize services right. Services is a cost center for us. What we want to do is put out all the information about how to have a great Mattermost instance, how to scale it, how to do really well.

We want to put that in the community for absolutely free, we want to make it web-discoverable. We’ve got these large forms, you have any Q&A, any problem you have with Mattermost just web search it and we want you to find the answer.

So that’s the investment cycle it’s: If you have a question coming from a customer, that’s fantastic. Go to the forums, or go to the documentation, add the answer in and send a link so that the answer becomes web-discoverable and the whole world gets that benefit.

So that’s the cycle we have, where we have a customer, we do answer their questions, we try to put that information out as quickly as we can onto the world so that Mattermost becomes easier and easier to install, easier easier to extend, easier and easier to customize.

So what we’re focused on is selling the commercial version of our Enterprise product to many many very large companies that can benefit the most from it. And then we want to give away the rest, to the rest of the world. That Enterprise business funds the innovation, and the support, and the growth of the open source project.

Community Support?

Mike Schwartz: Do you actually answer community support questions?

Ian Tien: Yeah, we’re absolutely on the forums, all the time.

And what we try to do is we want to encourage the community to answer and what we’ll try to do is wait at least 24 hours to allow the community go in, you know, have their two cents in and take the first crack, and then we’ll come in with sort of more experts.

But we really want to enable the community to grow and to build. And not only, you know, not only parts of the open source committee but also our partners who are experts at implementing and delivering the Mattermost and helping customers; we want to empower them to answer more questions, and so to get known in the community, and to participate and to enjoy.

And then behind them we want, we have our core contributors that will also support and answer.

Levels Of Support

Mike Schwartz: For commercial customers is there an SLA, or do they get quicker support?

Ian Tien: So on the forums no one really knows that you’re a commercial customer. People just ask questions and we try to treat everyone sort of the same.

What the commercial customers get is support through other channels that they can contact directly and we can be much tighter with them, depending on the support level.

So for very large Enterprises we’ll have our own Mattermost channels where we can go back and forth, and it’s very quick, and easy to share information. And it’s really embedded with us.

Mike Schwartz: So you’re using Mattermost to support your customers.

Ian Tien: Oh, absolutely. So our desktop app can actually connect to Mattermost, multiple Mattermost servers.

And we have customers that, you know, they’ve got their Mattermost instance in one tab and they have us in another tab. They can just switch back and forth and talk to us and go to instance and it’s great because they can share feedback you’re like – oh check this out, we built this thing and it’s amazing, and you know, we can pass it on we can share it, share the ideas.

Or there’s something they want to tweak in that user interface or there’s, there’s something we need to change, and we get that feedback right away.

And we have a monthly release cycle, so every month on the 16th of the month, we release a new version. You know, it’s a single Linux binary so very straightforward upgrade, and that’s how quickly we can react to feedback in the community every month.

Challenge Of Real Time Support.

Mike Schwartz: Is that hard, having customers be able to sort of ask you questions in real time, and sort of expect immediate feedback?

Ian Tien: It’s a mix, yeah.

So what we find is most of our customers when they’re onboarding, there, you know, our support is really around configuration right, so that they’re like okay, I can do high availability, I need to install this, and install this. It’s a lot less break-fix. So when Mattermost runs you know it’s sort of coming and going. So what we find it’s a lot more configuration.

So, yeah you’re right, like sometimes expectations of certain instant response. That said, a lot of our organizations are actually using messaging because they’re asynchronous.

Because they don’t want to use like a Skype for Business or something that’s ephemeral, where people come in a different time zone, they log in, there’s no history, they don’t know what happened between the whole day. With Mattermost it can be both synchronous and asynchronous, and people sort of do them in different ways.

I think working with our customers they kind of get a sense of the rhythm in the pattern. And it’s really, this is a very, very top tier support; these are strategic customers that even like co-develop with us. So they’ll be contributing code, they’ll be giving us really insightful advice.

So they’re really a part of the community, they almost feel like part of our extended team, because we’re all working together to make a fantastic product and to really develop it. It’s not very common and it’s not a huge number of customers we have at this top-tier support, but when it works, it works really well.

Mike Schwartz: Using open source as a distribution channel people, search probably for “open source Slack” or “open source Skype,”

Distribution Channels

Mike Schwartz: Using open source as a distribution channel people, search probably for “open source Slack” or “open source Skype,” and they find you, so I’m sure that’s an important distribution channel. But are there other distribution channels?

Ian Tien: Open sources of is a wonderful channel, and what we think about it, is sort of a, a growing series of channels that sort of escalate upwards.

So the open source is one piece, it makes it really easy to find, web searches from Mattermost keep going up, we get a lot of organic traffic.

The next tier is really around resellers. So if you’re look on our website under partners, you’ll find a listing of resellers around the world that can bring Mattermost and explain the product to more Enterprises.

So they might not be familiar with open source, they will understand the benefits of having a collaboration tool that’s completely under IT control that’s easy use, easy to install, so a reseller can go in and they can discuss the product in French, in Korean, in German and really

help explain to these prospects and customers our value proposition. And because we have this commercial product, it’s much more familiar as a commercial partnership, how to distribute.

So we get the benefit of open source and we get the benefit of traditional resellers and channel.

More Channel Details.

Mike Schwartz: Just curious – have you looked at the percentage of sales from channels, or how’s it sort of stacked up to your expectations?

Ian Tien: We’re really excited about channel. And it’s about, sort of how the story begins.

And how the story begins is there’s a lot of interest from Enterprises and the challenges procurement. And when you have a channel partner like SHI or Dimension Data, they’re plugged into a huge number to Enterprises, so we can do procurement through them.

It’s a win-win because those resale partners give the Enterprise the benefit of consolidated billing right, and really easy procurement.

We get the benefit of be able to fulfill the order; and when you have an ecosystem where everyone’s winning that’s very scalable. So it starts with just the authorized resellers and it’s like hey help us pass the paper, that’s what they do, and we have that partnership.

Then we’re getting into sort of value-added resellers. So companies like AdPhineas over in Europe, where they’ll provide local language support in local time zone. And we have a partnership to take the lead that we get in their region and work together to expand the business.

So again it’s win-win with value-added resellers. And then we have distributors like a FedResults, which is part of Carahsoft, in the federal government, and they’ll do even more.

So they’ll have a marketing organization that we can partner with to find opportunities in their existing customer base and really tell them about the benefits of Mattermost. In addition of that, they hold our GSA schedule, so when we sell Southwest Federal Government, we have a partner that really understands how to procure, how to work with these organizations, and navigate the system. So again it’s win-win.

So it’s all about creating community where your economics and your success is aligned.


Mike Schwartz: You mentioned that Mattermost is open core. So you have a free Community Edition and then you have other versions I guess of the software with additional features.

I’m just curious is the main revenue streams from the business license of the non-open core, or are there any other revenue streams, training, support, or is it just all hundred percent license?

Ian Tien: The commercial part of the business is focused on selling commercial licenses of our add-on’s that are built on top of the open source project.

So if we’re selling Premier Support, that’s only available to customers who are buying the commercial version. Because that’s our focus.

Now if someone wants to start a business supporting the open source version, that’s wonderful we’re not going to compete with them. We’re only going to provide services and training to the people who are using a commercial version.

Mike Schwartz: In terms of like percentage of revenue is there an 80/20 rule, where 80% of the revenue would you say is licensed? Or do some of the other like support contracts on license, do any of them make up more than 20% of revenues?

Ian Tien: So everything we do is subscription. There’s a portion for licensing and there’s sort of an add-on for that support.

We try to make our customer success organizations sort of cost neutral, right. So you buy the support add-on and what you get is that the extra layer support and you know, that revenue covers the costs of, supporting these Enterprise customers. Where we grow the company is really from the subscription revenue of the commercial add-on’s we do.

License And Renewals

Mike Schwartz: By using the software they’re required to keep renewing their subscriptions. How have you found getting renewals and making sure you retain the customers?

Ian Tien: Yes, it’s a great question. There’s a couple of things to talk about there.

Number one, just because how we started the product, and started the company, and started the open source project, we really didn’t like being locked into a vendor-hosted solution that had all of our data, that we didn’t have any choice.

So purposely because of that, when you use Mattermost, you have a hundred percent access to your data to MySQL or PostgreS[QL] database, and if use the commercial version and you decide, you know what, I want to use the open source version, all you have to do is take out the license key, and you got the same functionality as the open source version without the commercial add-on’s.

We make it that simple so no one ever feels like they’re locked in.

That said, renewals are really strong because people have validated that this product already brings value, they’ve used the open source version of for a while typically before they go to the Enterprise version. So they know it works, they get the Enterprise version for the compliance features.

So the renewal question is do I still need those compliance features. And for large Enterprises who are running these in the data center, it’s a pretty straightforward decision.

Mike Schwartz: But you have to chase them down, they, you know, didn’t sign your subscription, or we didn’t see that check, or – how is that process going? Because that can be tricky.

Ian Tien: Yes, so we have a back-office person that does that, so they’ll send out the emails, and you can renew using credit card, you can you use a bank, or wire in different currencies; so we support all that the back end.

Mike Schwartz: But does the license shut them down if they have it renewed? So is there license enforcement?

Ian Tien: So before a license fee expires, there’s a series of warnings that the system administration will get sort of in the interface, like hey, you have this many days before it expires. We’re very open with the sort of notifications you get in the grace period afterwards, and there’s sort of some slight changes in behavior afterwards.

What we do want is everyone to feel safe and comfortable running the software, and the license renewal is more about the procurement system then sort of like IT issues with backend.

Typically if you’re Enterprise you want to renew properly, get everything buttoned up properly, so it’s been pretty smooth.

Value Proposition

Mike Schwartz: To switch tracks a little bit and talk about the value proposition for Mattermost?

Ian Tien: So one thing we really think about is privacy-conscious Enterprises. And for privacy-conscious Enterprises Mattermost is the team collaboration solution that let’s them increase productivity and have this modern world of collaboration while meeting their custom security and sovereignty needs.

So there’s going to be requirements that these large Enterprises have to be able to be in complete control of their collaboration system. To have no third-party monitoring. To have many security options because it works with their private network, because they can do intrusion detection, because they can analyze every line of the source code, and their information security team can go over everything with a fine-tooth comb. And on top of that there’s custom security, custom compliance, custom auditing.

For those Enterprises that really value privacy, autonomy, control, and extensibility. And be able to go further than they could with any vendor hosted solution, any vendor hosted multi-tenant solution – those are our customers.

And you can read about that sort of positioning in, 451 Research has a note on us, Gartner in their workstream collaboration space has covered us, we’re a vendor to watch in the latest market trends report. We’re actually being named a Cool Vendor by them.

And it’s really about differentiation. It’s really saying – hey there’s a lot of collaboration solutions out there for Enterprises that are really privacy-conscious, that really want control, that want that layered extensibility, who want to have the option to go further in there customization of the collaboration solution.

That’s our differentiation. That’s where we really live.


Mike Schwartz: We talked a little bit about channel partners and, those are important to every business, but are there other partnerships?

Ian Tien: Yes, absolutely. When you run an open source project you have community on your mind all the time.

So Mattermost has over a thousand people who contributed to the open source project. On top of that there’s over a thousand open source projects based on Mattermost. We’re in 15 different languages. There’s a huge community extensions around us.

So for the open source piece, that’s growing really well, and that’s part of the engine that drives awareness and discovery, and more customers.

In addition to that, we’ve got our experts community in our channel. So how do we help distribute, how do we help deliver this offer into Enterprises, through resale and through services.

One part of that new community that we’re really excited about is Atlassian channel partners.

So as HipChat has announced its retirement, for these high-security companies that really want to do collaboration, especially if they’re Atlassian users, they no longer have that option now that HipChat is retiring.

So Mattermost is really stepping in, we want to make users of the Atlassian server products, that need total control, that need high security, that need to stay in private clouds or data centers, we’re filling that gap. And we’re working with Atlassian partners like cPrime, the largest Atlassian channel partner in the United States, to go meet the needs of those Atlassian customers.

And we’re working on new partnerships all around the world with the Atlassian channel to make their customers successful.

In addition to that, there’s certain communities from our customers that have built up around the Mattermost product, around the open source version, and eventually the commercial version.

One of the best examples there is NH-ISAC, which is a global association of information security professionals from the healthcare industry. So they’ve been running Mattermost for their community as an open source solution for awhile. We’re about to announce a partnership with them.

Where they’ll be using the Enterprise version, and we’ll have an increasing presents working with NH-ISAC, attending their summits, getting to know their community, the challenges their communities have around how to collaborate as information security professionals on a platform that is vetted and is under their complete control.

Challenges Of Open Source

Mike Schwartz: One of the questions I had was – you know open source has a lot of great qualities – collaborative development, and certainly for marketing, but have you discovered any challenges? Has it, have there been any negatives or any, has it hindered you at all along the way?

Was there any unexpected surprises, where you thought open source would be great but it turns out it was actually maybe sort of more challenging than you thought?

Ian Tien: Absolutely. Coming from a design background, we thought really carefully about what is Mattermost Team Edition, which is our open source version, and was is our Enterprise Edition.

So our Team Edition is kind of like a small office: it’s going to be really easy to start, you just step in there, everything works.

You got emojis, emoji reactions, you can do channels, you got GIFs, you have mobile apps and desktop apps, and it just, it works. And if you’re coming from Slack, it’s really easy, the keyboard shortcuts from Slack are supported, webhook integration from Slack are supported, slash commands; it’s very familiar.

Now we have our own additions. Like the ability to use you know, spaces, and non-English letters, and channel names, and we’re able to use hashtags and some other things like that.

The open source Edition which is for teams, is really designed for for that world, right. I don’t have a lot of ask permission for anyone, I can kind of do whatever I want, I trust everyone to be good citizens in this Team Edition.

And then for Enterprise, you’ve got all the complex things. You’ve got roll-based permissions, very sophisticated high-availability, all the compliance stuff. And we thought that delineation would work really well.

What’s happen is, people using the open source Edition are too often unaware there’s an Enterprise Edition.

So they take the Team Edition, which is built for small teams, and we have Enterprises that have hundreds and thousands of people use the open source version. And when we talk to them, they’re like, you know what you should add, you should add a permission system. But we have a permission system in the Enterprise Edition that they never sort of discovered. So all the problems that they have, that they think they have, are actually solved in the commercial version.

We just haven’t provide the awareness. That’s something we think about all the time do we want to, in our open source Edition, make it more obvious that there are these features for Enterprises, and they are in the commercial version.

And that’s a line we have to be careful about because, like well, do we really want to market inside an open source project?

And the consequence of not doing that is you have a lot of people using open source version who would easily upgrade if they knew they could just phone procurement and say, I want these extra features, they’ll make my life so much more productive. And we got like three thousand people on this instance – yes like, just go make it a piece of our company infrastructure.

So that discoverability is a question for us, and the consequence of not making the commercial version more discoverable in the open source project is brand. It’s people feel like wow, Mattermost can’t do these things – when we can. But then how do we let people become aware of it.

Gauging Community Size

Mike Schwartz: Do you collect statistics back from the products that you know who’s using it? Or are there any type of opt-in’s to send back marketing data? Or how do you gauge the size of the community?

Ian Tien: So any sort of telemetry you put into a high security product is never going to tell you the accurate story.

We had a lot of metrics looking at what’s our search volume – how many people are coming to the website, and were they unique visitors, and how many people are downloading. And we try to gather a lot of sort of anonymous statistics around this to sort of see the velocity and the direction of the interest, to make sure it’s always growing.

In terms of our actual user and customer base we want to make it super easy for them to sort of discover us and learn about us and contact us. And that’s part of the customer journey.

So they’ll come, they’ll find us, they’ll download. Somewhere along the way they’ll re-engage. And then we can continue that journey as they contact us, as we understand how they’re using us, and how we can add value.

So yes, because we’re dealing with privacy conscious Enterprises there’s a little bit of a gap in our story. What really matters is the flow of the start to the middle, so we can not only convert them over to the commercial version that they’re looking for but make them super successful afterwards.


Mike Schwartz: What are your thoughts about funding and open source, and is that something that Mattermost is seeking? You know, you’re in the heart of VC Mecca here in Palo Alto.

Ian Tien: So if you look at our website you can actually see a number of our investors. My personal view is that you know, finding the right investors is an incredible benefit for any organization.

So being able to have more capital means you can do more. You can accomplish more. We can hire more people, we can pay more people to build the open source project, to help scale the commercial business and we can go faster.

All we have in this world this time and if we can bring in capital to move things more quickly, I think as an open source project the most important thing is you’re open, and you’re transparent.

You’re saying like, this is what we do, we’re open core, here’s a commercial version, here’s our open source version, this is, this is what we do. And if investors are aligned with that vision,

and we’re aligned with those investors and what their expectations are, then it goes back to that great ecosystem of line incentives.

Our investors make introductions for us, they give us great advice and we’re a lot further along because we had that collaboration and help.

Closing Advice

Mike Schwartz: So if you are an entrepreneur who wants to use open source as part of your business model, do you have any advice for that person?

Ian Tien: I think the most important thing is to focus on the value you want to create. You want to build something that people love.

So start with – I want to build something that people love, and think about roles that open source can play: in your vision for the product, your distribution model, the community you want to build, and the business you want to build.

Mike Schwartz: Well that was really fascinating. Thank you so much for making time Ian, and best of luck with Mattermost.

Ian Tien: This is fantastic.

Mike Schwartz: That’s it for episode 4. Special thanks to the Linux Journal for co-sponsoring this podcast.

To the All Things Open conference for helping us to publicize the launch.

Music from Broke for Free, Chris Zabriskie and Lee Rosevere.

Production assistance from Natalie Lowe. Operational support from William Lowe. Thanks for the staff of Mattermost for logistical support.

Next week we’ll talk to one of the visionaries who inspired this podcast, Mike Olson from Cloudera. Don’t miss it.

Transcription and episode audio can be found on opensourceunderdogs.com. Until then, thanks for listening.

Episode 3: MariaDB – Open Source RDBMS with Michael Howard

Michael Howard is the CEO of MariaDB, one of the fastest growing open source database platforms in the world. MariaDB powers applications at companies like Google, Wikipedia, Verizon, Deutsche Bank, Telefónica and more. In this episode, Michael describes how the “Business Source License” enables MariaDB to monetize software and reap the benefits of an open source community and development methodology.



Michael Schwartz: Welcome to Open Source Underdogs, the podcast where we bring you the wisdom of open source business leaders.

Today my guest is Michael Howard, the CEO of MariaDB Corporation, the company behind the fastest-growing open source database in the world.

MariaDB is used by more than 12 million developers worldwide and is the database of choice to power applications at companies like HP, ServiceNow, and Wikipedia.

Michael, thank you for taking the time to join us today.

Michael Howard: You’re welcome.

Michael Schwartz: Why don’t we start the podcast with just a brief introduction of how you happen to join MariaDB and maybe a little history about MariaDB.

MariaDB History

Michael Howard: So MySQL was named after Monty Widenius’s daughter.

Her name was My, something like that in Finnish or Swedish. And Maria is his second daughter. So that’s the naming convention here, it is through Monty’s daughters. And as a result of course, the group of engineers that build MySQL and MariaDB are very attached to their family.

And when MySQL was initially bought by Sun, and then ultimately segwayed through to Oracle, there was a rift in the family. And in about 2009 there was a fork of the MySQL project and that fork became MariaDB.

And then some years later in 2013 MariaDB amalgamated, or put together, all the different aspects of the programming group the foundation, a support organization called SkySQL and in late 2013 that’s the MariaDB Corporation as you know it now.

Michael Schwartz: How did Michael Howard end up joining MariaDB?

Michael Howard: Michael Howard in the late, all through the 90s and late 90s was a disbeliever of MySQL and anything open source.

So that’s the origin of my story, in that, I could never imagine a group of programmers building a general-purpose database that would be able to tolerate an accommodate and support the primary functions of a business.

And I truly believed that the only way that that could be done is assembling very straightforward proprietary approach and let’s say it, my PhD thesis of that came when I joined Oracle in the 90s, and that’s what I believed.

But then as the world changed, and this is after 2000, you could see that the world was dividing. There were a couple of important signs of that and that was the Hadoop or HTFS world coming to play. It was the NoSQL databases coming into play starting with the caching layer.

And my view changed along with that new ark of software. So I knew the guys at MySQL and MariaDB for many years, didn’t think that they could bring themselves forward into true OLTP enterprise fortitude workloads.

But the world has changed. And I changed with it.

Michael Schwartz: OLTP – Online Transaction Processing?

Michael Howard: Yeah, you got it. Should have said that, but yes.


Michael Schwartz: So tell me a little bit about today. Who are the customers of MariaDB?

Michael Howard: Almost every company in the world now is touching MariaDB in some kind of way, in that we have displaced MySQL as the viable default relational database in all the Linux distributions. So even mom-and-pop shops that use Linux automatically have access to MariaDB, we have an addition that goes in those Linux distributions.

Somewhere between 12 and 60 million customers are touching MariaDB today through this unique distribution. And this is by the way one of the unique things about open source, that your ability to propagate and touch people across the world is enormous.

But that’s how it’s happening with the Linux distributions, and so to with the clouds in that, since it’s a default mechanism in Linux suddenly becomes a default mechanism in many clouds.

So we span from small shops that you’ve never heard of to the Fortune 10. You could look at things like AT&T, Verizon, Barclays, New York Mellon Bank, Development Bank of Singapore, Alibaba, ServiceNow, I mean it just goes from the most consequential companies in the world, even running airports. It’s quite remarkable.

80/20 Rule For Customers

Michael Schwartz: Just to drill down a little bit. Is there an 80/20 rule? Who are the most important customers in terms of Revenue?

Michael Howard: That’s a matter of opinion, how you do the 80/20 on that.

Because one of the problems that, let’s say the Hadoop family has via Cloudera or Hortonworks, is that Hadoop has become kind of part of the 1%, not big part of the 99%. And for me coming from proprietary areas you don’t want to relegate a company like MariaDB to the top 1% or 10%.

Therefore saying that your top 10% or biggest customers are the most important because here we want, and I’m going to use the word extremely, an extremely well-rounded business model.

So for example, in our world about 40% of our transactions on a quarterly basis comes from mom-and-pop shops. They don’t really think too hard about databases, but they kind of go with what’s easy for them to include as part of their environment. Then there’s another 40 odd percent where they’re making more strategic decisions but it’s still appended to a mentality of, for example replacing MySQL.

And then if we use those percentages there, there’s another 10 to 15% which is big time, and literally replacing out the likes of DB2, SQL server or Oracle itself.

And so I guess you could apply the 80/20 rule in each of those categories, or sale segmentations. But you couldn’t come up with an 80/20 rule across the board.

So there is sale segmentation, if you look at our documentation for example, it’s free. So anybody who doesn’t have a support contract can use a magnificent amount of content. But if someone has a production application and they need some sort of identification that’s a different level of interaction with the company.

We can’t talk to all customers, it’s impossible. It’s a challenge that cannot be overcome by people. If you look at it from let’s say, a business perspective like margins and productivity per person, there has to be more automation.

There has to be more automation in terms of buying contracts or buying things through eCommerce.

We even have to have easier ways where people can instantiate and test out new features instead of going it on their own, so we’ll do some things in the cloud to facilitate that.

So there’s many levels of interaction that must take place. But I think what we need to do to improve, really big time, I mean we have some major steps to make to be able to handle the numbers that we’re talking about, and we’re not doing that right now.

You know you look at some of the big players like Amazon or Netflix or whatever, they are doing it. So that’s the funny thing we’re a small company, under 200 people but we have the same level of interactions as the cloud platforms do.

Because, and sometimes even more, because MariaDB is on all those clouds. It’s a fixture there, so it kind of gets multiplied many times.


Michael Schwartz: What are the most important revenue streams? Is it license? Is it support?

Michael Howard: Well for me coming from the proprietary world and the SaaS cloud, I’m very formalistic about this.

So the most important from a just formalistic point of view is ARR, Annual Recurring Revenue, on a quarterly basis the amount of new annual contract revenue.

Things like TCV which is equivalent to bookings are less important to me, more important for let’s say my CFO who wants to make sure we have a certain amount of cash in our bank accounts.

So the most important thing is ARR, that’s the first level. Second level is what does that ARR constitute.

And again coming from where I came from, where the multipliers on services and manpower support is not as powerful, not as valuable to, let’s say listing the product at NASDAQ or having some big liquidity event in the future, the most potent value is product; license on product.

And that becomes a very interesting and almost somewhat contradictory or paradoxical conversation when it regards open source. So our subscriptions are the most important part, and those subscriptions come with many facets.

It comes with technology that cannot be procured or accessed in a general sense by just the Community Edition.

We save some stuff that’s private. It may or may not be licensed under something that we came up with which we believe in, very much not only for MariaDB, but for open source entrepreneurs world over, and that’s something called the business source license, which we should probably talk about.

So you have technology. You have things that are not necessarily available through the Community Edition. You have legal protections, I named one, in some cases limited indemnification issues come up in especially regulated industries for production applications.

Then of course you have the typical element which is, a typical element of open source which is support.

So that’s how we make money. Our revenue come solely from our subscriptions which has a spectrum of things in it to make it valuable.

Open Source Dilemma

Michael Schwartz: There’s a saying called the open source dilemma where customers only renew support contracts when they need help, when they have problems. How do you get customers to renew and retain them to get that revenue to be actually recurring?

Michael Howard: Right, there’s a… It’s an interesting topic because the good and the bad of open source is that it compels the engineer or the firm to build new things. So for example, in our latest release we have temporal processing, inversion tables, very sophisticated ways of supporting governance applications.

This is a very very, it’s a multi-dimensional data object, being able to be joined with a million other multidimensional data objects, the math is extraordinary; be able to do that in an optimized way.

When you put out features like that and you continually do so, you will compel, behoove, motivate very easily customers to continue with their contracts. So the virtuous cycle is inherent in your ability to innovate.


Michael Schwartz: Changing topics a little bit. Are there Partners who help you get MariaDB out there, or that you rely on for success? What is the channel?

Michael Howard: We do have channel partners but it’s not as mature as, let’s say older firms.

I think we’re just at the very beginning of it. And channel partners can come from an enormous number of different areas, for example it could be a cloud vendor that is actually a channel partner, right. Being able to serve a certain constituency, as you put it, through their Cloud. It could be your typical channel partner that has some kind of ecosystem, could be that.

The one that comes to mind, just kind of new, both I think for the world and for us is you take something like ServiceNow, where they’re serving 40% of the Fortune 2000. And many of those companies don’t even know that MariaDB is the engine behind every single workflow that is going on in those companies, which is quite extraordinary.

But in some cases they need to have on-prem support from us and so suddenly you have the world of cloud interacting with the world of on-premise and that becomes a channel play that you would have never envisioned before.

Value Prop

Michael Schwartz: Why do people use MariaDB?

Michael Howard: I’ll give you an example.

Let’s say you’re a big company and have a huge server farm, and you want to start looking into ARM as your chip bases. You’re not going to be able to use a process space database on an ARM chip, it just won’t be economical.

Maria happens to be thread-based and the pooling resources for threads on ARM is as good as typical non-ARM chips.

So it could be that you are trying to have more compact space for your server farms, less real estate, less use of electricity, less heat. It could be that you realize after 20 years that you’re paying 60 times more than you should for basic SQL standards.

It could be that the basic philosophy of the company that you’re using is so out of touch with what a modern system is that you want to move forward with your own stack.

There’s an extraordinary number of reasons.

I mean the the primary one is that when you go into MariaDB you’re not only partnering with one of the best engineering firms in the world. You’re partnering with other great engineering firms in the world that are contributing to the MariaDB kernel. And you know that MariaDB will never be out of touch because of the extraordinary open source community that is contributing to it and there is no equivalent.

At Oracle everything goes through Larry; he is the Editor in Chief of Oracle.

Here you have a much more diverse set of curators.

MariaDB Foundation

Michael Schwartz: One of the things I wanted to ask you about was the MariaDB Foundation. It’s the ideal of a lot of Open Source projects.

Maybe you could talk a little bit about how that came about and what their role is.

Michael Howard: So the way I look at the foundation is they are the steward of the project’s repository. And they are the ones that teach people how to contribute to that repository.

Allowing freedom and equality for creativity, innovation through code, persisted in repository that is formalized under the foundation. And they’re advocacy to developers to help with that, their outreach to Linux distros. That’s the kernel of what they do.

Michael Schwartz: Where’s the line between the company and the foundation in terms of a decisions?

Michael Howard: Well I mean, there are simple lines, I’ll start with the simple ones.

They don’t do any support, you can’t go to the foundation and say I have a problem with something or bug, they won’t, they don’t have anything to do with that. That’s where the corporation comes in.

If there is a hot patch that needs to be applied because of a security issue, they’re not, they don’t act in that manner, they don’t respond that way; the corporation can.

Obviously you can’t get any legal privileges through the foundation. Let’s say you have a big idea that you want to put into the project, but it’s maybe a little bit over your head to really do it in the right way. You know just because there’s an open repository doesn’t mean that you can willy-nilly put something in there that’s mediocre.

But there are certain projects, I’ll give you an example: The Development Bank of Singapore wanted MariaDB to have a PL/SQL Oracle compatibility layer. That’s an extraordinary difficult project and could never be done themselves. And that’s when you enlist, for example, our engineers in the corporation.

Foundation Feelings

Michael Schwartz: So not to put you on the spot..

Michael Howard: You have so many times already!

Michael Schwartz: But nonprofits always involve politics and so how do you feel about how has it gone?

Michael Howard: I think it has pros and cons. I don’t think it’s a clean deal by any means.

I think that when you have independent entities they each have their own egos and they each have their own aspirations to spread their own wings. You know it’s just like when two people are discussing a subject, each one of them kind of wants to have their opinion supersede the other.

So it’s part of humanity to have that innate conflict but also innate need to work together.

Business Source License

Michael Schwartz: You mentioned license before, and I didn’t want to make this podcast a deep dive on license because you could probably do a podcast by itself just on the licensing part. But you did mention the business source license.

Maybe you could talk a little bit about what is the business source license and how does it relate to the other licenses perhaps at some of the other products or offerings come under?

Michael Howard: I’m going to start in a strange fashion on this narrative and that is: When I announced business source license I should’ve had a set of security guards from the open source community. I wasn’t sure what might happen to me.

Because it was disagreeable to them in some emotional ways in that, there are a lot of people out there that are pure open source zealots, and it’s an ideology like any political or kind of religion that can get the best of certain people.

But I felt that the company needed to be able to have as much financial resources as possible to make our products great.

And since we were stepping into the enterprise zone, which meant that we had to simulate or actually replicate many things that SQL Server via Microsoft or Oracle through Oracle does for their customers. And financial resources are extremely important to do that. And database features as you know, it’s not like a SaaS application.

Sometimes it takes many years to build one feature and those people are dedicated, like the version tables in the temporal processing that I mentioned, that was a very difficult feature, it took us years.

So how do you reconcile the world of free and the world of financial resources? How does an entrepreneur like yourself go to a VC and expect any funding if there is no business model. And at the same time if your whole company pivots off open source, again how do you do this?

After some deep thought the business source license seemed to be the best balance that I could find.

And what it is, is essentially very simply postponed open source. You get one of our products that uses the business source license, and for a certain set of time that particular version you pay for.

But at a certain point in time let’s say 36 months later it becomes open source. You can self support, you can do anything you want to it, and you’re not dependent on the corporation.

So it’s just trying to bootstrap product so it can meet the expectations, especially when these expectations have been built over 20 years with many billions of dollars.

Michael Schwartz: Sort of like a patent that expires, maybe.

Michael Howard: Yeah. That expires, that’s a good one.

Except I guess the deviation would be that even though the patent in this case is on for that amount of time, you can look at the code, it’s completely transparent just like open source.

Go to repository, our repository, it’s on GitHub, look at branch, you see it, you can contribute to it, you can do all those things. There’s even certain times when the license says you can use it for free, you know that kind of thing.

But, if you going to into like a production mode, and this is the case with a MaxScale, our database proxy. If you have many nodes that you’re serving with this proxy – you got to pay for it at some point. But that version will under the file name go GPL at a specific date.

License Enforcement

Michael Schwartz: What are your thoughts on enforcement? Is it on the honor system?

Michael Howard: So, right now it’s honor system. I think we’ll have to do some kind of home calls at some point and identifiers.

So there are times when certain customers try to take advantage. Where they’ll buy let’s say one or two copies of the paid version, but really the problem is emanating from thousands of free things that they’re using.

So right now it’s the honor system. There’s been some transgressions that I’ve seen. And of course the bigger the company you go to the less transgressions there happens to be. It’s more in the lower end where I see abuse.

Advice To Startups

Michael Schwartz: Any advice for entrepreneurs who are starting, who want to use open source as part of their business?

Michael Howard: So my advice is: It depends on what part of the stack you are.

So if you are in infrastructure you have no choice but to be somewhat commodity based or open source based. I don’t really believe that security for example has to be open source.

I see, you know I have a little security background. So listening to you, there was no reason to make, let’s say FireEye, open source. Just no reason, no one would expect it.

I think it’s interesting that you guys have gone in that area a little bit but the key management systems that I work with I never expect them to be open source.

On the SaaS level, you know you have one interesting juxtaposition between SugarCRM in Salesforce right. Has Sugar really gotten major benefit from its open source disposition?

You know I think what’s happened is the proprietary vendors have milked the industry so badly, that the time came when that practice had to stop. And MariaDB is sort of a symbol of that arrest.

Michael Schwartz: If you’re an entrepreneur, you’re saying if it’s infrastructure you should go open source.

Infrastructure & Open Source

Michael Howard: You don’t have a choice anymore.

Michael Schwartz: You have to, okay, that’s interesting.

Michael Howard: You really don’t. I mean you know, you have Intel maybe coming up with a chipset that’s not open source or their 3D XPoint, you know, that’s not open source.

But if you look at, you know let’s say a no more commodity-based analogs to that; they’re going to end up winning. I mean Intel will lose again because they’re charging too much.

So infrastructure I don’t think there’s a choice.


Michael Schwartz: Any thoughts about how do you stay true to the open source mission that maybe has helped build the culture around the company, and still fund the company, or build a business?

Michael Howard: Well, I think you see with very few exceptions VC’s funding open source companies. I think the market has already made that decision.

I mean there’s just no comparison between the funding percentages for open source versus lets say closed. There’s just no comparison. So I think the decision has already been made.

So then the question is if you have an idea at store at a storage software level, let’s say, or a container, you have no choice to be open source.

So then there are two things that come to mind. One is, you have to employ something like business source license and that’s the only way you’re going to get funded when you are presenting to a venture capitalist.

You have to have some way of bootstrapping your company with some revenue, a support meal ticket just does not make sense from a margin perspective, or the multipliers that VC’s are expecting in a liquidity event. So it’s dead on arrival if you go that way.

So it’s very simple you can’t just go open source, you have to have a business model behind it that makes sense. There’s only a couple ways you can do that.

Michael Schwartz: Thank you so much Michael for sharing your wisdom and best of luck to MariaDB.

Michael Howard: And to you too at Gluu.

Michael Schwartz: Thank you.

Michael Howard: Alright, bye.

Michael Schwartz: That’s it for episode 3. Transcription and episode audio can be found on opensourceunderdogs.com

Special thanks to the Linux Journal for co-sponsoring this podcast, to The All Things Open Conference for helping us publicize the launch.

Music from Broke for Free, by Chris Zabriskie and Lee Rosevere.

Production assistance from Natalie Lowe. operational support from William Lowe. Thanks for the Staff of MariaDB for logistical support.

Next week we’ll talk to one of the youngest startups in our series – Ian Tien, CEO and Founder of Mattermost, one of the leading chat and communication platforms.

Until then, thanks for listening.

Episode 2: Netgate – Secure Networking Software with Jamie Thompson

Jamie Thompson is the President of Netgate, the company behind the popular open source firewall project, pfSense. Both Jamie and her husband Jim, who is also featured on this episode, have spent the last 15+ years making secure, high-performance network connectivity tools available to the masses. In this episode, Jamie and Jim discuss bootstrapping their company and Netgate’s unique business model that involves monetizing hardware that implements their open source software.


Michael Schwartz: Welcome back to open source underdogs, the podcast where we interview leaders from successful open source software companies and bring their stories to you.

Today my guests are Jamie and Jim Thompson, founders of Netgate, the company behind the popular open source firewall project pfSense which is over 1 million installations.

Jamie could you start by telling us a little about Netgate and your journey founding it?

Founder Story

Jaimie Thompson: Sure. Netgate is a open source network security company, both contribute to and utilize open source software and create products from those projects.

About me personally, I’m too old. I’m from Oklahoma originally. I was peak woman, there was a peak year for women in computer science and that was the year I graduated with a degree in computer science. And that was 1984, a long time ago. And I have a master’s degree in applied cognition and neuroscience which is essentially experimental psychology.

Worked as a software engineer. Early my career switched over to doing more of the sales engineering, worked for Sun Microsystems. I’ve worked for here in Austin a company called Tivoli Systems, which was ultimately went public in the 90s and got bought by IBM.

And I guess we started Netgate in about 2002 and I’ve been doing it ever since.

Jim Thompson: So Netgate was originally the name of a firewall that we’d done.

The first time we had an open source company called SmallWorks here in Austin that started with open-source I think my first open-source contribution was a port of new E-Maxx to a convex supercomputer in like 1987.

And there I went to Sun and that’s where I and Jamie met. Building out their worldwide network and got pretty involved in network security, that point I download the first set of proxies. Around 1989 or so I did that, because we had a need internally.

Once we started hooking up all the sales offices and remote facilities to this internal WAN and everybody wanted access to what we now know as the internet. It was really more ARPANET then. So I built a proxy so that people could get out and go fetch the latest version of the X11 system or something like that.

And that sort of, I saw that and saw people use it and sort of identified a need for actually agreeing to build a packet filter for the Sun Systems that was sort of the origination of the original Netgate packet filter or firewall.

Jamie Thompson: So when we started, Jim came up with this packet filtering firewall back in the early 1990s. And it was also open source but in kind of a different way – when you bought it from us, we gave you the source code so that you could analyze it and make sure that there weren’t any backdoors and you couldn’t sell it further on.

But you could certainly take it and analyze it internally, and in fact one of our early customers was Wells Fargo Bank. So we’ve kind of been dealing with network security and firewalls and took a detour through wireless for a decade. And have always been interested in and involved in network security and how do you create both privacy and security, both for companies and for individuals.


Michael Schwartz: Who are the customers of Netgate today?

Jamie Thompson: Well it’s really interesting, the software is very lightweight on providing any sort of feedback for us. It is open source, you can download it off the internet and run it.

You don’t have to sign up for anything, we don’t create any not paywall, but there’s no registration wall, to get the software. So we don’t always know who it is unless they are either asking us questions, either on the forum or in support or if they need some professional services or they want a private port or something like that.

So we’re we actually are learning more and more about who the customer base is at this time, but we have everything from individuals who are running it in their home labs, and then those people sometimes will take it into their companies.

And we also have a lot of government entities who run it. And then there’s actually in entire governments that run it, not the United States government but portions of it actually do, yeah. So it is used both inside the US and then other government entities around the world. Lots of individual, yeah, lots of S&B. There are some service providers who also use it.

It’s also integrated into some other products, so you’ll see people who’re providing some sort of service and they use our software either for the transport or for a firewalling, that sort of thing.


Michael Schwartz: Is there software that people can buy? Or do individuals just download the open source?

Jamie Thompson: Well both. They actually can download it for free or we also take that software and we package it onto hardware.

So Netgate utilizes the project and creates a product from it. And then the product is packaged either with hardware or in the cloud and we sell support and services around that. So really how we’re…

Jim Thompson: Some part of that is system integration. You know, product test and release testing. that kind of thing. We test primarily on the things we sell.

So there is a lot of system integration or ports to get on new hardware, so we expend the effort to get on some of them and make that experience. You know really clean and easy. But then we have to maintain that going forward.

So we test and release in part just on what we sell, we make a generic version available to anyone. But the actual, if you will, hard work of making sure that things really work has to be done on some hardware platform.

This we have to focus on, we sell or cloud platforms, virtualization, that kind of thing.

Support Services

Michael Schwartz: Do you sell support services?

Jamie Thompson: Not everyone who purchases hardware from us gets a support contract or needs professional services, a lot of people already know how to use the software. It’s straightforward enough that they don’t need any extra work.

So like a lot of other companies out there, we can’t sustain the project on just support and services.

Jim Thompson: There are both large and small customers on the forum. US Army Cyber School we discovered on Reddit. Literally, one day. They were already a customer and they are user of pfSense. We didn’t know it because we have no way of testing them.

So again, it’s sort of across the entire range of customer types or segments of the market, if you will. We only know and work with people who have approached us.

Now, we know the approximate attach rate for people that we sell to versus how many instances of pfSense are turning up in the world. We can’t measure on a very granular, non-granular or scale that we see, more and more IP addresses if you will, looking for updates, or downloading additional functionality via packages, that kind of thing.

So we can see in approximation of the size of the installed base on any given month, and we know who we sold. And you know that ratio is about 1 and 20.


Michael Schwartz: Are there some companies who are OEMing the product?

Jamie Thompson: Well, there are companies that have taken it – it is open source, and so they can integrate that into their own product or in the cloud.

We have versions of the software available on the public Cloud so they can utilize it and in their businesses or integrated as part of their own software as a service product offering.


Michael Schwartz: How about channels? Do you mostly sell direct, or do you have any integration partners who help you reach customers?

Jamie Thompson: We have both a set of hardware resellers. So there are people who wanted to be involved with pfSense and especially in other countries, or in other languages we don’t have the ability to speak with them directly, we have ported the, it’s not call port it’s a…

Jim Thompson: Endemic translations.

Jamie Thompson: Yeah we’ve done a bunch of language translations, do you remember how many?

Jim Thompson: It’s approximately 30.

Jamie Thompson: Yeah I was thinking 30 as well. So the software runs in these other languages and obviously we can’t speak all those languages, no one here is that good.

So we have in-country partners who can either sell support service, and we have both people who are doing straight reselling, we have people who are MSP’s or MSSP’s. We have people who are buying it direct and then daily lease it. So there’s all kinds of different models.

We also, in terms of channels on the cloud, we do support some of the cloud service providers. People who do the integrations to basically create an entire stack. So if you wanted to move your application or the set of applications to have as a company into the cloud you could do that and the cloud service providers, their partners would help you migrate all of your data and put together your infrastructure and help you monitor and manage it.

And our best partners right now are in Europe. Some of them again are MSP’s or MSSP’s and some of them are just selling direct or helping people integrate.

Value Prop

Michael Schwartz: What’s the value proposition of Netgate?

Jim Thompson: We are the developers of pfSense primarily. There are community contributions that you can actually go out and look at like the list of Git commits. And you filter out the ones that you know staff here have done. It’s you know 90% plus Netgate.

And so since we’re the developer a lot of people feel safer when they’re dealing with the people who actually make software, you know, why wouldn’t we give the best possible experience on things we sell.

Jamie Thompson: We employ a lot of engineers, we have people who have commitments for FreeBSD. We have people who are just fantastically gifted both in hardware and software. We couldn’t do it without the engineering team, they’re just fantastic.

And it’s, you know the kind of thing you can’t get from a community because it’s hard to prioritize that open source development over what you’re doing in order to be able to be paid so that you can eat, and have a house, and all those good things.

So we actually do employ a huge team of Engineers to work on this open source product. And do the testing and the architecture and we run the infrastructure and…

Jim Thompson: You’d be amazed at how low-level we have to go to fix some issues. You’re talking about, you know, peak of second timing on NMMC bus, or we run into in situations where Ethernet PHYs don’t have the right choke on them, and so you can see more power in the PHY than you otherwise would. And we’ve had to actually go back and debug designs we didn’t do.

So there are there literally three people on staff who are EE’s, know them and engaged with EE on a day-to-day basis. The 4th who knows enough about it he could, he used to design medical devices. So we have a depth here, the technical depth that some people would find astonishing.

Use Of Open Source

Michael Schwartz: Has open source served as the main distribution channel for Netgate?

Jim Thompson: Well principally it’s a licensing structure. It’s almost all open source licenses are based on copyright, the original, the ability of the original author for assigning to control that copyright or leverage it.

The original, if you will, of the GPL turned copyright on its head that’s why some people call it copyleft. They use copyright to enforce the fact that you can control the copying. It’s an interesting hack if you own the legal system.

We have benefited enormously as has, you know, RedHat and a slew of other companies you can name from, if you will, marketing via these you know user groups. They’re in control of their own conversation, there’s no marketing conversation in the room. There’s all this experimentation if you will, 10000 flowers.

Try the successful ones are successful, and the unsuccessful ones you never heard about. So yeah we’ve engaged that too, and the fact that the the software is open source and free has been quite a bit of the actual marketing.

People discovered it and use for their own purposes and introduce to people who we were never gonna here from unless they approach us directly. It spread largely through that if you will, word of mouth or word of forum, association of use pfSense, it’s great.

You can still see that going on today. There are hundreds and thousands of people I will never meet who are using software we create.


Michael Schwartz: What is the primary activity of Netgate?

Jamie Thompson: Software development is the primary activity. If you say what are you as a company? We would say that we’re a software company.

But in order to leverage that software, in order to be able to actually continue to develop that software we do need an income stream, we need a revenue stream. And since we are self-funded, we don’t have VC, we’ve had to do that by selling hardware and services and renting it in the cloud.

So just like any any other company would do, I’d say that our number 2 activity actually is probably hardware development, which is kind of weird.

We have hardware, we work with some of the ODM’s and OEM’s on hardware designs and so we’ll go to them and say well we need, an Intel box that has 4+ next, you know i350 and two i210’s, don’t want those i211’s.

So like Jim said earlier, we have a lot of hardware depth that you wouldn’t expect for a software company. And that really isn’t normal, I think that, you know, if you look back we’re kind of more of a throwback to the kinds of companies that you had in the 80s where you would, you would develop both the hardware and software together and sell tell the system.

So that’s what we’re doing, we’re selling appliances and we’re providing both the software and the hardware that it runs the best on.

Jim Thompson: Vertically integrated.

Jamie Thompson: Yes, oh there you go.

How Did You Figure Out How To Vertically Integrate And Self-Fund?

Michael Schwartz: Normally I would think a hardware company needs a lot of capital. How did you figure out how to vertically integrate and self-fund?

Jim Thompson: A lot of credit card debt.

Jamie Thompson: Yes there’s been a lot of credit card debt, that’s true. When we started down this path with the hardware we were taking components that were available, and we were basically integrating those components and we were doing it on a pretty small scale.

And as we were able to convince people that we knew what we were doing in terms of both the software and the hardware and the integration. And we just had some really good solid products out there. That, actually, stuff that we sold 8 years ago is still on the field running today. And we basically just slowly grew and slowly grew.

Then we would reach kind of the limit of that hardware or of the supplier. We had one supplier, so we were doing 500 a month and we’re like okay we need to go 600 a month. Well we can’t do that, we can’t manufacture that much.

Okay we have to go off and find a different manufacturing and we’ve slowly gotten to the point where it’s just been a pretty much continuous ramp to the point now where we can go to some of the larger people in China or Taiwan or wherever they happen to be. And say, okay we will sign up to this many systems over this period of time and …

Jim Thompson: We can write large PO’s and back them with financial strength of what we built together over the past decade.

But has taken reinvesting almost everything we made to continue to make the business grow. So we live, sleep, and breathe Netgate… While raising a child!

What’s Next For Netgate?

Jamie Thompson: Kind of more interesting thing is that over time we’re looking at the software which actually started with Manual Casper, manu wall back in the day. And we’ve improved the software, we’ve ported it as FreeBSD has changed, we gone along with that.

One of the big things that we’ve just done is we’ve all ported to PHP 7.2. So we were running with the older version of PHP and the current person that’ll be coming out here shortly we’ve had to to update to 7.2 because of course you know software end-of-life’s, over period of time it goes to a no-support model.

One of the interesting things we did about three and a half years ago, four years ago, is we sat down, we said you know – if we were going to rewrite this what would we do? Because the questions that were being asked and the requests we were getting from the customer base who talk to us are like you know, it’s easy to do one firewall.

It’s easy for me to sit down in my home lab and control the firewall for my house. But at work I’ve got 50 of these things and I have to go control them individually just have a GUI, rather than a GUI, do you have an API? Can we use rest RESTCONF for NETCONF, or basically as architectures have changed and people’s thinking about management has changed.

We saw that there really wasn’t a way for us to change that software because of the way that was originally architected. If would be really difficult to put an API in. So while we’re also moving pfSense along we’re also thinking in the back of our heads what do we need to be doing differently.

We have actually some new software and it’s also based on open source, it’s Linux-based. It’s called TNSR.

And TNSR is really the result of a lot of architecture discussions within the engineering group and looking out to see what other people are doing now in terms of managing their networks and providing security for their networks.

So now you’re looking at orchestration and automation. A lot of people are are looking at well how does this run with Kubernetes how does this run with containers.

How I manage 50 of these things at once, and so not only are we working with pfSense but we’re also working with TNSR is really, well it’s available today in the cloud, but it’s really trying to kind of solve the next generation of problems around network security.

Why TNSR Was Needed

Jim Thompson: In those environments computers and the networks they run on have disappeared inside a machine, there is no box, there’s just one box, it’s the big box that you don’t even know it or see it.

It’s on a cloud provider somewhere, so on Amazon or Google cloud or Azure or something, and you never see it, and you never see the networks around. You just have the sort of remote ability to control via an API so there’s no there’s no blinking lights, there’s nothing left to touch.

And so that ends up being very different environment, these Kubernetes environments have messaging rates sometimes that are you know hundred thousand messages a second you have to be able to tell what’s happening.

So TNSR, which Jamie started talking about, is sort of an answer to what’s starting to occur in-network.
How Much Profit Gets Re-Invested In R&D?

Michael Schwartz: How much profit gets reinvested in R&D?

Jamie Thompson: All of it, basically. Yeah.

So, basically as we’ve grown the company and have been able to add more people, or we’re really lucky and in the respect that we took whatever we had, and said okay, what’s the next thing that we can do to make ourselves be more useful to our customer base or to be more useful to enterprise.

One year we added 24/7 support. And suddenly, oh okay, you guys are real because you do 24/7 support now.

But we’re no more real the next day then we were the day before. It’s just suddenly people felt more comfortable because they can get a hold of us and they can we can help him walk through their problems. And of course no firewall problem ever occurs at noon on Monday it’s always midnight on Friday.

So going to the 24/7 support model was huge for us, starting a partner program, a worldwide program, to help support people in-country and in-language with our partners was another huge bump for us because we’re now able to actually answer the questions that people had in a way that was more comfortable for them.

As things have changed out if to focus more on the cloud and to focus on this automation and orchestration.

Non-Integrator Partnerships

Michael Schwartz: What are some of the non-integrator partnerships?

Jamie Thompson: We didn’t see a way for us to take the software that we had and move it forward so we started looking around.

We have supported the BSD Foundation in the past. We’ve also now joined Linux Foundation and we’re big contributors with LFN, with Fido. We also support Clixon. So, not only we are supporting the software that we have written, we’re also supporting groups and other software projects where we’re either contributing with monetary contribution or code.

So you actually if you got look at at Fido, you’ll see that we are in the top five maybe even top three contributors for VPP which is the vector packet processing, that LFN is doing.

So we’re trying to help contribute to the ecosystem not only the code that we’re writing and that we’re putting forward, but helping, trying to help move networking and security long a little bit further in ways that we couldn’t do on our own.

But we can we can contribute to it, we can help.

Jim Thompson: In the same way that we have a community for pfSense were part of other communities around some of these constituent technologies. There’s even marketing on that side as well, the Linux Foundation markets LFN.

Fido’s BPP is a big part of that. So we know we show up at conferences and occasionally talk and participate in various mailing lists, you know online conference calls about where should we go next and that kind of thing. So all that tend to count for contribution.

Making the technology move along even though that technology ends up in potentially competitors product system.

Like I said, in the same way we have a community of people that we support. We’re also constituents in a community of some of these technologies that we take advantage of.

VC Money Risks

Michael Schwartz: You’ve avoided raising venture capital, why?

Jim Thompson: As with almost any decision there are pros and cons to that particular decision. The money is what people tend to focus on, how do I get this thing bootstrapped to get enough people working on it while I have something.

I’ve been the CTO of two companies, that combined I helped raise over a quarter billion dollars of venture capital, over seven or eight rounds. One of those companies was in product and one of those companies was in services.

My best advice is while things are going according to plan, the venture capitalist are here to help. They will open the Rolodex and they’ll introduce you to additional partners, give you some legitimacy into accounts that potentially you didn’t, you couldn’t enter.

If you can’t execute on the plan that you told him about, that’s when things can take a turn.

Jamie Thompson: That’s when they change the plan. Sometimes.

Jim Thompson: You know, VC’s are financial engineers. They’re there to optimize return for their fund.

And so the best advice I have for anyone who’s looking at this is to go out and really understand what VC is and how it can help. And what the VC’s goals are versus what your goals are. And if you can find a way to line those it can be great.

If those goals aren’t aligned then it’s just money and there are potential other places to make that happen.

Jamie Thompson: If it’s something you can articulate and something that is a physical product you can always go Kickstarter, or crowdfund, or there’s all kinds of other ways to do it.

But like Jim said I mean we’ve both been involved in companies that have been VC-backed and we’ve seen them sling the company around to try to meet whatever their goals are.

And so that’s actually one of the things that we’ve talked about occasionally, is okay are we at the point now where we can’t continue to bootstrap it ourselves, are we at the point where we should take some money and so we can do some additional marketing.

And we were real hesitant on that. Trying to decide.

How do we move forward and how do we maintain the spirit and the feeling of the company. How do we maintain our open source roots because we were all very focused on security and privacy and we believe that everybody should be able to have security and privacy if they want.

But if a VC makes it, if a VC says okay well now you have to get everybody to register, well you know we’ve got people who don’t want to do that.

Startup Advice

Michael Schwartz: Any advice for entrepreneurs starting an open source software business?

Jamie Thompson: It’s harder than it looks. It can be fun, it can be difficult.

But if you have what they call today grit, if you have that ability to stick with your idea and keep going but yet, if someone is able to influence you and you change your mind help steer you in a new direction, that’s great.

But it really, it’s on you and you find, they would say find your passion, but find a thing that’s really interesting to you and stick with it.

Jim Thompson: Businesses survive at the will of their customers. Solving customer problems and providing value to the customer is literally why you have a business. Because without that nobody is sending money your way unless they feel bad for you.

As an entrepreneur you have to be willing to engage with and talk to your customers and prospects. You have to be willing to take the call that says your code is broken, fix it now. You have to be able to hear rejection when somebody chooses another solution and get up the next morning and go back to work.

These things all sounds trite but a lot of people the computer field are into the computer field because they like working with computers, not other people. They have friends but population at large can be scary to approach because what if they say no.

The best thing you can do for yourself as a business or as an entrepreneur is to be willing to interact with the people who think your product is interesting. And they would like to find a way to use it and they have potentially other ideas about how it could be used and you have to be willing to listen.

If the number of bright people on a given body of people is, I think as Bill Joy famously said, is the log of the number of people in a group, so the number of smart people in a group is log of number of people in a group, and he didn’t say what the base was, but it’s this diminishing returns, a larger group you have the fewer smarter people you have per out of basis.

But these ideas occur everywhere. And open source is really one of the answers to the question of how do we adopt other people’s ideas. Somebody can have an idea and they can develop an open source project and it flourishes or it doesn’t. If it flourishes it was a great idea, incorporate that into their technology stack.

It’s the same thing with customers, some customers will help you focus your products in ways that you weren’t going to think of. And so that ability to be open to hear both positive and negative messages about the thing you’re doing with your life or that part of your life is really critical.

Michael Schwartz: Jamie and Jim, thank you so much for sharing your insights and best of luck with Netgate.

Jamie Thompson: Thanks for having us.

Jim Thompson: Thank you.


Michael Schwartz: That’s it for episode 2. Transcription and episode audio can be found on opensourceunderdogs.com.

Special thanks to the Linux kernel for co-sponsoring this podcast, to the all things open conference for helping us publicize the launch.

Music from Broke for Free, Chris Zabriskie and Lee Rosevere.

Production assistance from Natalie Lowe. Operational support from William Lowe. Thanks for the staff of Netgate for logistical support.

Next week we’ll talk to one of the superstars of open source business, Michael Howard CEO of Maria DB. Until then, thanks for listening.

Episode 1: Anaconda – Python Data Science Platform with Peter Wang

Peter Wang is the Co-founder and CTO at Anaconda, Inc. Anaconda is the world’s most popular Python data science platform and is the industry standard for data scientists. In this episode, Peter discusses Anaconda’s fast growing business, and how they use enterprise tools to monetize their open source software.


Founder Story

Michael Schwartz: For our introductory episode of Open Source Underdogs I’m sitting here with Peter Wang, CTO of Anaconda, a distribution of Python math libraries used by data scientist all over the world.

Peter could you start by telling us about your background and how Anaconda got started?

Peter Wang: My background is actually in physics and I end up going to software after college because I really like coding. And I sort of found my way to doing consulting using the Python scientific DAC if you will, that was kind of in the early and mid-2000s and as I did more work in that area I realize that there were significant commercial adoption opportunities for Python.

But the scientific stack was you know, it was still coming together at the time, right, there is some significant gaps and difficulties and how to use it. Then once I saw the Big Data wave starting to hit I recognized that enterprise data analytics would be something that Python be very good at, and people already starting to self service and use Python for.

But there were gaps, and so we started actually as Continuum analytics in order to promote the use of Python for data science and data analytics beyond just scientific computing that we want to do some innovative projects to fill some gaps and holes, things like that. And then also to build a enduring and sustaining open source software company.


Michael Schwartz: Who are the customers of Anaconda?

Peter Wang: Our customers are large medium sized companies that buy our enterprise software, so that is a machine learning and AI enablement platform. It’s a very different pile of bits the that most people in the open source world who download free Anaconda are used to thinking about and so that’s you know the Enterprise customers buy that in order to support their data scientists that are using Anaconda internally in their businesses.

Value Prop

Michael Schwartz: What’s the value proposition of Anaconda?

Peter Wang: When you look at what data scientists need to do there’s a lot of data they need access to, there’s a lot of compute they need to run their stuff on. They need to do a lot of different things that actually traditional Enterprise IT is not very well equipped to support them in.

So our Enterprise platform comes in, and gives IT an easy way to provide a governed and secure computer environment for data scientists. They can collaborate, build models, build notebooks, build interactive dashboards, gives data scientist an easy way to also deploy those in a production way.

Now in businesses lot of data science can put toys together, they can’t it for them to then throw it over the fence to get IT to make something that actually runs for other people’s business to use, that can take a very long time.

So the Enterprise platform eases the deployment, it gives IT a way to manage and to govern the kind of data science it’s done, gives data scientists access to packages and software internally inside the enterprise firewall. And then for people who have only lived in the open source world that may sound really strange but lot of people they live in corporate environments that are very logged down, they can’t get access to software.

So with Anaconda Enterprise we provide a way for IT to feel good about there being a vendor to provide the software and the data scientists are happy because they get the latest and greatest versions of all the packages they want.

Challenges Of Open Source

Michael Schwartz: What have been some of the challenges of using open source as part of the business model?

Peter Wang: So one of the things is that we both use open source in building our Enterprise product, we also, what Enterprise product does is makes open source available for people to use, right, so there’s things like machine learning libraries like scikit-learn that we don’t write and we provide. So there’s challenges associated with simply providing open source to Enterprise users.

Then there’s additional challenges of just using open source software in our proprietary enterprise software, right, there’s sort of two sets of challenges. With the former, with the distribution open source software I would say that most businesses are actually starting to understand that open source software is a key part of the software development stack.

For machine learning and AI applications in particular, it’s an integral part of it, you can’t do without using open source software, so Enterprise IT is starting a clue into the fact.

The challenges run into are that even though they are clued into this, there’s still procurement and legal and security and governance questionnaires that come down to us that sometimes you’re just like WTF, you look at this like you know like there’s no way we would make reps and warranties on a piece open source software that has global contributors, right, or yeah we have customers asking for unlimited indemnity of something or the other.

It’s like we don’t even write half of the software, it’s your users internally that want to use it. You know, we’re just making this available for you.

So there’s like some of those kinds of, I would say disconnects and impedance mismatches between enterprise expectations.

The biggest challenge I would say is that enterprises, they think of data analytics providers and vendors like SaaS or someone like that, that comes in with a big giant piece of shrink wrap, very expensive shrink wrap software.

That’s all just that vendors bits and so they’re used interacting with vendors in that way. So when we come in and we say we’re a vendor that’s an enablement platform, we have our pile of bits, we indemnify and support and provide you warranties, but then part of that is we provide fluid rapid access to a ton of additional capability in the open source ecosystem.

They really have a hard time, it’s really legal oftentimes that has the challenge of separating, you know, the bits that the vendor wrote and the bits that the vendor’s providing access to. They have a really hard time disambiguating those two sometimes.

Anaconda Platform

Michael Schwartz: Can you talk a little bit more about what makes or what defines a platform and what does that mean exactly to you?

Peter Wang: So for me a platform is something that, it’s like a market right and so it facilitates an interaction between providers of value and consumers or users of value and it makes it from an M times N problem to a smaller like M plus N kind of problem.

So in our case, we have you know, there’s hundreds and hundreds of developers of open source libraries that want people to use their software, and that you know there’s a lot of capability there. And then there’s thousands, tens of thousands, millions of people inside businesses that want to use the software. But they don’t know which software they should use, they don’t know what software is secure to run, what versions are the most up-to-date, things like that.

Our platform essentially is a bridge between the open source innovation space and development around these machine learning libraries to the Enterprise, very governed environments, in which the software needs to run.

So creating a place where people can bring their goods so to speak, so in this case software packages and capabilities, notebooks, it could be datasets, whatever. And then where the consumers of those things can come and actually pick what they want, that’s what makes it a platform.

So we ourselves of course we built quite a bit of functionality into the platform but really letting this additional generated activity happen inside it.

It’s not merely, oh here’s a big pile of software and then we have some extension points so that people to plug-in things, it’s not a plug-in ecosystem.

It really is a platform that provides people with this bridge that lets people go both ways.

Cloud Offering

Michael Schwartz: On your website I noticed Anaconda.Org mentions a cloud offering. Is that a direction that you think is is promising? Or is the main focus the enterprise software suite?

Peter Wang: I’ve always been very interested in data science in the cloud.

Actually one of our first offerings as a company in 2012 was a cloud-based on demand and notebook computation system called Wakari. And then we eventually shut it down because it was a very hard way to make money at the time and it takes a lot of capital to actually get a software-as-a-service company going.

And so right now Anaconda Cloud as it exists is mostly a place for people to host packages and notebooks and to share those with people but we don’t actually execute those notebooks on people’s behalf.

In the future I cannot make any promises about our long term product roadmap, but I do feel like cloud-based execution of notebooks and of models, things like that, it’s a natural need that emerges and you know we may do something in that space.

What’s Next

Michael Schwartz: What are the most promising areas that you’re investing in or looking to innovate in, in the future?

Peter Wang: Well, you know, the space moves so quickly and it’s evolving quite a bit so we really are looking at I think for most Enterprise software companies would be, because of a very short time windows, or innovation sort of new, de novo innovation work.

We’re mostly right now investing in the existing projects that we have. We’re also investing, there’s a new project we just released for data access, to really solve that data reproducibility and data sharing problem; lightweight data catalog and data access library called intake that we are very excited about.

But moving forward we will be looking at doing, you know more things with regard to model management and helping people share models in a more reproducible and seamless way. And you know beyond that it really depends on how the business evolves.


Michael Schwartz: So of those different areas where you could generate revenue, which one is generating the most revenue as a percentage? Is it license?

Peter Wang: We’re primarily, the vast majority of our revenue comes from software licenses.

That’s wasn’t always the case, we actually made a pretty dramatic change over the last year. Flipped almost completely 180° around from being mostly consulting with some small amount of, relatively small amount of software revenue to the other way around. Where it’s mostly software revenue and a much smaller amount of consulting, training, services, things like that.

License Model

Michael Schwartz: Can you share any details about how you went about licensing?

Peter Wang: We start as a company to support and promote the growth of Python for data science.

So we’re something that’s very much of a community, right, it’s not like Travis and I sat down and said: Oh, aha, we have this genius idea for some crazy new cool technology, we build it and then we’ve vend it.

It’s, we’re of this community, we built more innovations to help, amplify the the efforts of the community, we’re always of the Python data science community. So, the ethos of that community is to license everything permissibly so MIT, BSD licensed, LGPL sometimes.

And so it wasn’t really feasible to do any kind of open core business model around that. Some communities, R as different right, the R Data Science community, R is all GPL based. And so classically there have been two companies, one got acquired by Microsoft, that were heavily doing work in R around open core model.

In the Python world I don’t feel like the community has an appetite for open core business models. And so we very quickly came to understand either A) we have to sell something directly to the community so our users and they’re our customers.

Or, we figure out some kind of enterprise software that we sell, that is a proxy need, fills the proxy need or as a proxy to the open source of free stuff that we give away.

So the business model we converged on was essentially that. We give away this stuff you know, you give away burgers and you sell the Coca-Cola or something like that, right.

It’s not, you get a free burger but if you want two patties, or if you want the burger with cheese then you pay extra, right. That’s not the model we have, so Anaconda, all the work I do are around Anaconda and the packages the distribution of the testing with all that stuff give away for free.

We will give it away for free forever, because we know that when businesses start using this stuff more, and more, and more, it will drive proximate need for for other kinds of things. And the first significant proxy need that showed up was the need for a management platform or some way of managing and governing the use of open source data science packages inside the business then it became managing the collaboration around Jupiter notebooks and things like that.

And now lately it’s evolved much more into deploying models and looking how models are running, you know, the actual models in production is a much larger part of what’s driving business pain and so those are prominent, all three of those things I mentioned the repository package governance, the collaboration and then the management of models of production.

All three of them are core components of our platform offering in our Enterprise product.


Michael Schwartz: Would it be fair to describe that as tools around the open source?

Peter Wang: I don’t think it’s tools as much as it’s, I mean the nature of the software itself is to provide an infrastructure for you know spinning up servers, deploying things. It’s not, they’re not really tools as such.

I mean actually the end user experience, when they’re sitting there, like if you’re a business user, as a customer that’s bought Anaconda Enterprise, you’re using the browser login to our Enterprise platform and you get the the Jupiter notebook experience there.

You can also use Anaconda as you always used it on your laptop or something, and the Conda package manager will connect your internal package repository. So it’s not like you’re getting extra tools, you have the same tools you’ve always had from an actual data science day-to-day perspective.

But using them in an Enterprise environment is much easier for you now.


Michael Schwartz: Maybe switching gears a little bit to the marketing side. Does Anaconda just sell itself? Because the community has a strong underlying, you know science Python community? Or what are some of the channels that you use to get the software out there?

Peter Wang: Well, so the enterprise software – so very few things in the world sell themselves. And the reason is two-fold, one if there’s not much demand for what you’re doing, then you have to go and tell everyone about what you’re doing.

If there is a lot of demands what you’re doing you’re going to competitors that race in and provide messaging around why their product is better than yours, right.

So in both cases if there is or there isn’t a demand and you’ve got to be doing some kind of selling.

Now for us the the machine learning platform, data science platform, that term is now a term of trade, right, in the business so we have people at the CIO level, CTO level, that are like, okay we need a data science platform, right. They know what that is, they know it’s something they need to do to support their data scientist.

Now it does help us that those data scientist will tend to already be Anaconda users, we do have access to their mail list and social media and whatever else. We have a pretty visible, pretty well-known as a brand, so the brand everything is top-notch. We’re known for our investments, interaction with the open source community.

So all of that stuff helps us in the actual sales process. But in the end of the day we do have to do marketing, we do have to inform people about what the features are, why they would buy.

It’s like any other piece of software you’re trying to sell, there’s real activity you have to do there.


Michael Schwartz: So without going into actual prices, but maybe just around pricing philosophy – were you able to quickly figure out what your price should be? Have you changed a lot over time?

Peter Wang: It’s changed a lot of over time. Depending on who you’re selling to, it really, you know, we’ve always been in the enterprise sales sort of mode, for our support training, things like that, as well for our products.

So from the beginning we had this understanding that enterprise sales is a different kind of animal than you know consumer sales, setting prices and all those things.

There’s a whole process to sales and sales strategy, and neither my co-founder nor I would claim to be experts in that, although we have some perspective and some background in it. So it’s important to actually work with sales people who kind of know what they were doing. And we’ve been evolving overtime, we’ve been learning overtime, how that whole thing works.


Michael Schwartz: Maybe we can talk a little bit about partnerships. Do you have a strong partner network or who are the partners that you’ve built up over time that you think have been important to you, getting out there and getting to market?

Peter Wang: So Partnerships, we’ve always worked with some other companies, but in terms of actually driving to sales success, that’s been a fairly target activity over only the last couple of years. Because using partners as a channel – their sales people have to know how to sell your product, right.

They have to know how to close a deal and include your product in the pricing so you have to have really, really clear definition of what is the product, how to talk about it as an integrated part of the partner offering, all those different things. And I would say that’s again, that’s something we’ve only been doing relatively recently.

Prior to that our partnerships really consisted of technology partnerships and sort of marketing and exposure kinds of things.

I mean the data science field itself is also relatively new, at least in terms of enterprise adoption of this stuff. So we actually partner with, there’s sort of two classes of folks: One of them are hardware vendors, actually. So Intel and Nvidia Microsoft, the cloud vendors like Microsoft, Google, Amazon all that.

But the hardware vendors AMD, Intel, Nvidia and gosh I probably could use some others there, IBM obviously, they have hardware. So with those we are partnered on technology as well as some go-to-market activities.

So we have a compiler called Numba that have distributing computing system Dask. Those are very, very exciting for these hardware partners because they want to showcase their chips running at scale, they want to showcase the kind of performance wins.

Now if they require end-users to have to write extremely low-level code in order to get those wins it’s a non-starter, right. So the idea, it’s not just performance but it’s performance as accessible.

So using our tools and technologies they can get your average data scientist with not some like C++ optimization nerd. Your average data scientist can use some of our libraries and then be running these things you know and really showcase their performance.

That’s some harder partnerships, then the cloud partnerships as well as some others. They’re really around the data science tools, around notebooks or visualization. Some cases they’ve been with our enterprise platform as well so it really depends.

So like with Microsoft for instance, we are partnered multiple ways.

So our Anaconda distribution is installed on Azure, it’s available as the Python runtime inside sequel server itself which is pretty cool. It’s something you can install as part of the default Visual Studio installer if you say I’m data scientist, and you select that profile, it will install Anaconda for you. So that’s a really cool partnership.

And then on the flip side of it, we’ve included the VS code editor inside the Anaconda distribution installer so that that’s one of the editors and ID’s that’s available for end-user.

So there’s multiple aspects of Microsoft relationship, but those are probably the most significant ones. But these partnerships are very exciting for us, they’re still in the early stages, you know, right now we would love to see driving more revenue through those.

We are investing in those relationships because we believe we will be able to drive a lot more revenue through them.

Ecosystem Partnerships

Michael Schwartz: What about non-, let’s say business partnerships, any organizations, or foundations, or other communities that you found that had been really useful for you to promote the product?

Peter Wang: Oh yeah, well so we primarily stay in the data science space, right, in the Python ecosystem.

We’ve been growing our inclusion of the R ecosystem as well because many data scientist use both Python and R. And so we’re trying to find better ways to play nicely and be a good participant in that ecosystem.

But on the Python side we’re actually I would say pretty unique in that we started the company in January of 2012 and then by March we started the effort to create a non-profit to support and sustain several open source projects in the Python ecosystem.

That nonprofit is called NumFOCUS and that’s been a really good partnership of ours.

When I started the PyData sort of global community sort of effort, we attached that essentially to NumFOCUS as well. So now NumFOCUS runs as a center point of coordination for PyData meetups, PyData conferences all around the world.

It’s really grown, just really blown up. So that’s a constant and really great partnership with that foundation. That’s, I would say probably the primary one at this point.

There’s a lot of community events and things like that, that we do as well just to be good stewards you know, and show up. But those drive a lot of awareness for us.

One of the challenges from a business perspective is, if you say too close, just mingling with your friends, you’re not going to sell much because, in our case our friends are all users of the free stuff, right and they’re not really the people who are signing the procurement checks to go and buy the enterprise software. So yeah, that’s one of the dynamics there.

Keeping Current

Michael Schwartz: How do you stay current on what’s a very hard domain?

Peter Wang: Yeah it’s tough. I mean I don’t… I wouldn’t go so far as to say I am current. I am current on some things and there’s other things that I’m not as current on.

We have just a very, very broad range of technical expertise under one roof here. So we have people working on compilers, people working on distributed computing, people working to visualization.

People who are working on JavaScript front-end things for the enterprise software with people who are working the back-end. So many things – containers and all the back-end coordination orchestration kinds of things.

Our consultants and trainers who go out in the field and talk to customers they get exposure to a wide variety of different kinds of modeling problems, optimization problems, what’s the state-of-the-art in machine learning, in deep learning, all these things.

And I just have the great privilege of being able to talk to all these people and kind of glean from their learning as much as I can. Now I do some deep dives myself on some things, I have more time now than I’ve had in awhile. That’s part of it.

But, most recently I’ve actually spent the last I would say 3 or 4 months really nerding out on, believe it or not, the physics and the anthropology and the human ecology around open source and open source ecosystems, communities, and how we do sustainable open source; how we continue innovation while maintaining stable software.

The nature of software businesses as a whole, especially as it pertains to open source. So these are areas that I’ve been actually thinking about quite a lot and I found that my physics metaphors or my physics intuition has helped me a lot in thinking about the System Dynamics of that human ecology.

Open V. Commercial

Michael Schwartz: What are some of the challenges you think of starting an open source business or, I should say a business that uses open source as part of their model, versus straight commercial software company?

Peter Wang: I have a snarky answer – which is that the challenges that face an open source company are no, really, theoretically, no different than what face in a commercial company.

In practicality they’re different. Because a commercial company one tends to have a fairly reasonable expectation that the investors of the company and their shareholders are the ones that the company’s optimizing for.

If a company does a great job that’s measured very quantifiably in returns to shareholders. That’s the American model of capitalism, that’s what a company does. So if you screw up the company. The only people you’re going to piss off are your shareholders and your investors. Probably your employees too.

But in the open source side, open source software company there’s actually much more explicit or intentional vision that matters.

Most companies, they have a lot of like vision, mission, blah blah blah. For the most part once you get to a certain size employees are all like, yeah this is to make money. Unfortunately, you know, that’s just the reality of it. Now some businesses have highfalutin goals, most don’t actually even attempt to play to those.

But it open source companies the founders and the people who are stakeholders in it, there’s a broader set of stakeholders.

So who you piss off is a much bigger range than just the shareholders and employees. And so I would say that’s the thing – is that failure means different things.

And for open source companies usually – it’s sort of like you split the whole world of living things and prokaryotes and eukaryotes, right. Like in the world of companies, of open source companies, I would say you can cleave it in to those that do open source as a means to an end; and those for whom the open source bodies part of the ends itself.

And how open source embodies an ends – now that itself can vary. For some people there’s a religious belief that all software should be free and by golly, we’re going to build this thing to be free.

And those companies sometimes can struggle because there’s this reluctance to charge for software because you believe software is free, and so they are limited to a particular kind of growth curve because the corporate financing is available to them is of a particular stripe because they’re consulting or services company.

The ways that they can fuel innovation is also limited because they cannot go and sell a product as easily as someone else could.

So those for whom just open source because, it’s open source, those companies who have that ends, you know they’re kind of constraint to one part of the landscape.

Then there’s other companies for whom open source is an ends and a means and part of it is that it embodies some technical vision that the founders cared about. That technical vision is either only achievable through open source or gathered momentum and gathered user base, and a way of doing things, a technical perspective, that is now at this point so steeped – like the open source is part of its DNA – that it would seem like a violation and a rejection of everything that the soul of the company is, to walk away from that. That’s a different kind of ends, right.

And then there’s others like, I think for my perspective we’re little bit of that, but then also, from my perspective there is a fairly opinionated view for us that open source is, it’s a statement that were not going to lock people out by closing the software.

So it’s almost like a statement of – we may fight as a business and compete in a landscape that’s rich and filled with many kinds of competitors but one thing we’re not going to do, is we’re not going to use closing down access to software as a means of either charging rent or fighting dirty, or a way to charge rent so we don’t have to innovate anymore.

So in that sense being open source company means that you are essentially committed, there’s a covenant to your customers, to your users, to the ecosystem, to your employees. There’s a covenant that you are going to innovate, because it’s so much harder to be just a rent seeking monopolist as an open source company just around the software itself.

So, it has a bit farther range than you’ve expected from the answer, but…


Michael Schwartz: Actually, you brought up one thing I wanted to ask you. Does Anaconda actually have a lot of competitors?

Peter Wang: There are some things that we do that are unique in the world. So on certain aspects of the technology and some of our product features I would say we are very unique.

Now there are certain other things that we do, in particular the software that we sell that, you know we think we’re the best but there may be other companies who think they’re our competitors, right. And so I think that from that perspective, and certainly it’s up to the customers and not us, to say.

So there are times when customers will you know bank us off on other things and so we definitely have competitors in the enterprise space.

FOSS Development

Michael Schwartz: And you mentioned open source as requiring you to innovate. But has it also helped you to innovate?

Peter Wang: Software is a collaborative creative activity. And then some, for some problems, for some kinds of projects, the kind of open collaboration that open source represents are the most effective way to harness collective intelligence and get something really useful out.

In other cases open source is actually, the open source development methodology if you will, a way of engagement. It’s actually not necessarily the most effective to harnessing innovation.

And the dirty truth is if you look at the way the open source project actually roll. For all you know, all the Kumbaya aside, you know how they actually roll.

The most successful ones are launched by one, two, maybe three person founding team. I mean 3 is pretty rare, usually it’s one or two people, usually one actually, and they like a blaze of glory, they dropped this code in the world. And other people start glomming on and that initial nucleation site around the initial feature set, the collaborative dynamics of that. That really sets the future, like the seals the fate of the project a lot of times.

But usually open source projects they launch when it is a blaze of glory sort of innovative leap from one person’s brain.

And even now like, if you look at there’s a really sad or interesting way of looking at this, like the maintainers of some of the most critical open source projects in Python that are used daily by millions and millions of people that back billions of dollars of commercial infrastructure activity whether it’s power grid whether it’s keeping satellites up, whatever… I could fit all those maintainers in my minivan.

Now one can say, wow witness the amazing leveraging power of open source but you could also look at that say wow that’s really sad, we’re under supporting and investing these projects. So I am a big believer in open source, I don’t want this to sound like I’m like poo-pooing it, I’m a big believer in it but I’m also a realist.

And I think that open source in the early days had a lot more of the, you know, the OSI, FSF kind of days and all that.

I mean a stallman, coming out and planting a stake in the ground and saying we’re doing open source, the commercial Unix people can go you know, they can go pound sand, and we’re doing open source. That was important as a stake in the ground at the time, and over the 90’s as, you know, the Linux folks tried to educate everyone about, hey open source, okay here’s what it is, here’s free software, here’s open source.

Now, there’s a very different dynamic, as businesses are like, yeah we don’t care. Like the golden era of software is over, its services, it’s machine learning enabled services. These are where the top end of the value chain are.

There’s a real squishing compression dynamic on software that’s happening, specifically just on open source software. And if we’re not cognizant of that dynamic, if we don’t step up as fans of open source software who love the collaboration dynamics of the community, if we don’t step up and defend that and say actually, if you’re going to rely on this infrastructure you need to be paying for the maintenance of it and not rely on volunteer labor.

You know I think that’s a conversation that the free software world needs to have with the commercial world.

Venture Capital

Michael Schwartz: We found sometimes that being open source actually is a hindrance to us. Do you ever feel that, like that open source makes your life more difficult?

Peter Wang: Yeah. For venture capital fundraising, open source is a huge liability, absolutely. Because VC’s don’t understand open source. The big ones, I would say.

Like, smaller funds that have maybe a little bit more, that can take a bit more of an opinion, be a bit more of a gambler on business models, things like that. Maybe, certainly there’s angels, you know there are angel funds that they believe in the technology or something like that.

But the but the vast majority of VC’s, they’re there to really pattern match against known business models, known growth curves, you know things like that.

So if your business model is working to charge the software they’re going to ask, well how much value is the software? What do we believe the predictable growth curve looks like for this.

If your business model is we’re going to get users, they don’t care if give away the software, right.

So it’s all about, when you talk to the VC’s, how do you present them with articulation of your long-term value or not long-term value, but the returns you generate on their investment. That’s often the only conversation they really care to have.

So if they’re investing in software and you just giving it all away it literally looks like you’re just giving away their money.

In fact – if you were to say we’re going to take the money, buy iPhones and give it away to everyone, they would at least understand that a little better, because it’s something tangible in an iPhone right, they can say well we’re going to get on the other side of it.

You know, or you’re going to give away anything like MoviePass or whatever, I wanna give away free movie tickets to people and lose money on movie tickets. VC’s are not afraid to lose money, they want understand what that money is buying them. Right.

If it’s buying something else, some other number – eyeballs, or users, or something else, which you can then show how that converts in the long-term revenue, they are happy for you to give away the money.

But if it’s simply we’re funding software development and you’re giving it all the way, and you can’t show conversion that they believe in to revenue then it’s not going to fly.

VC Alignment With FOSS

Michael Schwartz: Were you ever concerned that perhaps the investment would come with strings that would make you to give up some of the mission or culture of open source?

Peter Wang: As we were talking to investors that was certainly one of the concerns.

I mean that with the reason we picked the investors we did was because we felt that they were mission-aligned and that was that’s a real luxury, we were very lucky in that.

And we worked pretty hard of the fundraising thing, especially as first time founders, not knowing what we’re doing. We talked to a lot of people. Probably screwed up a lot of meetings but, we ended up with some folks that actually really understood, not only the commercial potential of investing this, in the space in this company but also the mission, so we’ve been pretty lucky with that.

Again I would say that it’s rare to find that.


Michael Schwartz: What advice would you give someone who wants to start an open source software business today?

Peter Wang: Ultimately the act of going into business, it’s not separable question from why are you going to business at all, right?

Like if you’re going to start a business it’s because you either A) want to fund some activity; B) you believe you have some unique thing that you can sell to the world or some unique service you can provide to the world.

There’s actually very few, a small set of valid reasons for starting a business, in my view. And attached to each of those reasons, if you do open source as core activity the question you have to ask yourself is, am I doing this as a means to an end or am I doing this, is this one of the main things I’m starting business to do. If it’s the latter which I think is really where the meat of your question is, if someone says I want to do open source software but I also want to make money doing it somehow.

There’s many people who can, by building open source software, use it essentially as a marketing or brand awareness tool and then they can freelance, they can build a very healthy consultancy from getting their name out there and being known. That shouldn’t be downplayed as a valuable thing, and you can build a reasonably good small business around that.

But that’s not going to get you into unicorn billion dollar valuation territory, it just won’t, because to get to that level you have to somehow get to a certain size of revenue.

How do you take down certain amount of revenue, either you produce something like a piece of software that’s extremely valuable that you sell a lot of. Or you have a ton of people doing a lot of work.

Scaling up a consultancy is highly non-trivial, most software geeks who care a lot about technology are not wired to scale-up, a lot of meatware if you will. Once you get to a certain number of headcount you have got to figure out as an entrepreneur and as a technical entrepreneur how to let go of parts of it and let someone else actually help you grow that and that will change the culture and that will probably change part of the mission too.

So if you want to scale the certain revenue sizes you’ve got to do that either the basis of providing a really valuable, scalable service, or software.

I would just say understand the dynamics. Don’t blindly rush in with a whole bunch of optimism and then just curse your fate like, there’s a real dynamic here.

Investors coming in to help put more gas in your gas tank. They want to understand what roads you’re on, how far you’re going to go. If you can’t communicate to investors on the basis that they understand, about what your business model revenue models are, then you have no business asking for them for their money. You know, don’t get mad at them, right.

So, I mean it sucks in a sense that we live in this world where people do not think more about investing capital in socially-aware activities or in a generative sort of effective labor. That’s a broader conversation outside the scope of this interview probably, but that’s a systemic thing which I hope will resolve in 20-30 years time. But at this point of time it’s just the reality of the investment you know field, what it looks like.


Michael Schwartz: What were some of the businesses that you looked to as you were looking to scale Anaconda as businesses that you could model after?

Peter Wang: We recognized that actually because we were not doing open core that took a lot of our peer crowd of open source companies off the table, right. So like, MongoDB is an example.

Michael Schwartz: Could you just define open core real quick?

Peter Wang: It’s where the core of the software is open source and then to use, to put more data through it, or to run it on certain kinds of machines with more cores, or to do blah blah blah you know additional features and additional whatever, charge money for it.

Michael Schwartz: So like an enterprise version and a community version.

Peter Wang: Yeah, community version only supports 100 users, enterprise is unlimited. Community version can only run in the cloud, enterprise version can run on prem, like things like that, open core yeah, it’s best way to define it I guess.

Michael Schwartz: I see.

Peter Wang: The core of it is still open source, you know it’s still legit open source.

Dual licensing I think also falls under that. Right, so like you basically, the core is open source GPL and then commercial license. That’s that dual licensing is also something, something that people do.

So you can use it for free for to a non-commercial setting but as soon as you take it in the house in a commercial environment your lawyers and their total like allergy to GPL will cause you to go and give IT a call to go buy the commercial license of the software so you don’t have to risk, you know, virally contaminating your internal software with GPL. Those are the kinds of models, and so we felt like many things in the data analytic space were in that kind of model.

And certainly databases have this a bunch. We didn’t really, we didn’t see those as comparables. And when you look at platform software there’s RedHat. Really not that much else to compare to.

Now we look at some of the Java framework companies that manage to sell, some got acquired by RedHat, some got acquired by Oracle. For the most part I mean once we figured out that we’re just going to sell enterprise software, you know by seats, sometimes by notes to customers.

It actually became pretty straight-forward how to think about the business, so in that case it was really drive a lot of usage of the open source stuff and that’s going to drive a smaller but correlated usage and demand of the enterprise stuff, that was it.

Final Thoughts

Michael Schwartz: What did I miss? Or is there anything else you want to add?

Peter Wang: I would say that right now I sort of just glanced by this comment earlier but I really do think that right now the world of software is in transition, so people who want to start open source based software companies now should think long and hard about what is the value chain for software actually look like.

It may be better just to do a software-as-a-service kind of thing and then you get to own the customer relationship, you own much more the value chain, etc.

Now there’s downsides to that as well but that’s something to think about because I think that the, and Jason talks about software eating the world, and that may be true, but I think less and less of what it eats has really high caloric value right; more more the high calorie stuff is going to other kinds of things.

So I would encourage you think strong about what it means to be a software business in the modern-day, especially as things like Amazon, Google, Microsoft eat the world of cloud services.

The second thing is understand what it is about open source that you love, that’s intrinsic to your mission and figure out, be very, brutally honest with yourself about what of that you want to preserve and what are the right mechanisms for preserving that.

Don’t assume it’s just a money problem don’t assume it’s just, oh if I got out from this like horrible soul-crushing corporate job I could do open source all the time.

There’s always going to be a whole bunch of yak shaving from a business and management perspective no matter what you do. Understand why you’re taking that particular road if you’re gonna take that road.

And the third thing is open source doesn’t exist in vacuum. It exists in a human ecology of users, contributors, competitors, and evolving technology landscape.

You need to understand to think strongly about whether or not the thing you’re building has long-term sustaining ecology value and if it does then it’s worth investing. Otherwise you might want to think about how to pull various other pieces of your ecosystem together to something that’s more valuable as an agglomerate.

Michael Schwartz: Peter Wang, Founder of Anaconda, thank you so much for sharing your wisdom with us. Best of luck.

Peter Wang: Thank you very much, thank you.

Michael Schwartz: Well that’s it for this first edition of Open Source Underdogs. Special thanks to the Linux Journal for co-sponsoring this podcast.

To the All Things Open conference we’re launching on October 21st.

Music from Broke for Free, Chris Zabriskie and Lee Rosevere.

Production assistance from Natalie Lowe. Operational support from William Lowe. And from the staff at Anaconda.

Next week we’ll talk to Netgate, a rare bootstrapped open source company who’s also based in my hometown of Austin, Texas.

Until then, thanks for listening.

Episode 0: Intro and words from our sponsor, the Linux Journal

In this episode, we introduce the podcast and hear from Doc Searls, Chief Editor at the Linux Journal, the original magazine for the open source community, and one of the sponsors for this podcast.

Transcript coming soon!