natalie

Episode 21: Couchbase – NoSQL Engagement Database with Ravi Mayuram

Ravi Mayuram is the CTO and SVP of Engineering at Couchbase, a NoSQL vendor responsible for the Couchbase Server, an open source, NoSQL, document-oriented database. In this episode, Ravi discusses how an entrepreneur may successfully orient themselves around the problem they’re solving.

Transcript coming soon!

Episode 20: Redis Labs – Database for the Instant Experience with Ofer Bengal

Ofer Bengal is the Co-Founder and CEO of Redis Labs, home of Redis, one the world’s fastest instant experience databases. In this episode, Ofer discusses the evolution of open source software in the cloud-hosted market.

To note: This interview took place a few weeks prior to Redis Labs’ announcement of an updated license for modules created by the company. You can read more about the Redis Source Available License (RSAL). Open source Redis continues to be licensed under BSD.

Transcript

Intro

Michael Schwartz: Welcome back to Open Source Underdogs, the podcast where we cash the collective knowledge of founders, who use open source software as part of their business model.

No company has been more central to the discussion of open source business models than Redis. Just google Redis and Amazon, and you’ll get back pages of results.

When I was in Tel Aviv, it was a great honor to interview Ofer Bengal, CEO of Redis, in their awesome new office.

So, without further ado, let’s cut to the tape.

Thanks so much for joining us today.

Ofer Bengal: Thanks for having me.

Michael Schwartz: I’ve started four companies, which is more than anyone I know, but I think you’ve got me beat. How many companies have you started? And how did you come to get involved with Redis Labs?

Ofer Bengal: My story is a bit unusual because of my background, I’m an Aerospace Engineer. I got to the high-tech business by some sort of coincidence.

For years, I was designing airplane modifications and installations on war airplanes like F-16, F-15, etc. I did it for the Israeli Air Force, and then, I started my own sort of consulting company in the same area, doing work for American companies, Israeli companies, got bored with it, and started to invent toys. Sold some toy concepts to American companies like Coleco, the Time – I don’t know if you remember that company – the Cabbage Patch Kids, Hasbro, etc.

Then I wanted to start my own toy company, which would not just invent toys but also manufacture them. It was 1980 some, I prepared a very nice business plan to raise money for a toy company, can you believe it, so, no one wanted to invest in that company, except for one person that was from the high-tech business in Israel.

They were two brothers, who were the fathers of the pioneers of the high-tech business in Israel, their name is Zisapel. The guy told me, “give me the stuff to show my kids, if the kids like it, I’ll invest.” That’s the due diligence. He came back after the weekend, “the kids liked it, and I’m going to invest.”

A few days later, he said, “You know, I even have a better idea – why don’t we start a data communications company together? I’ll put the money and you will manage the company.” I said, “why are you even suggesting that because my mind is in toys now?” He said, “you look to me like an entrepreneur, pushy, etc, so I would like to bet on you.”

I said, “no, thank you, I’d like to carry on with toys.” But later on, once I couldn’t find more investors, I gave up, and I started a data communications company. That was back in 1991, or so.

And from then on, that company I took public at NASDAQ in ’97, and then later on, I started another company, which I sold, and then another company, so it became back door to the high-tech business.

Michael Schwartz: So, how many companies is that?

Ofer Bengal: Basically, in terms of starting companies, I started four companies in my career, and then managed another two on request of venture capital firms. One for Sequoia, one for Star Ventures. And these companies were a bit in distress, and I was called to try to save them.

Origin

Michael Schwartz: In 2015, Redis Labs hired Salvatore Sanfilippo, the original author of Redis, but before that, you were using the Redis name, you’d raised quite a bit of capital from Silicon Valley – did you learn anything from that process? And looking back, would you have done anything differently?

Ofer Bengal: Not really. Like many other companies, where you start is not where you end.

When we started in 2011, the database market was dominated by very large companies, like Oracle, and it didn’t make much sense to start a database company at that time.

We, me and my partner, who started the company, came from the application acceleration space, and we thought that we always knew that whatever you do in the front end of the application, the database is always the bottleneck. We thought that we should do something around it, and we started with cashing systems for accelerating databases.

At the same time, when we just started, we found a relatively new open source by the name of Redis, which was started by an Italian guy, Salvatore Sanfilippo. And we thought, hey, this is cool – let’s take it as the engine under the hood for what we were doing at that time. And so we did.

But after a year, we saw that this open source is becoming extremely popular, and we said, “maybe, we’re in the wrong business, and maybe we should turn the company and make it a Redis company.” So, we did that.

A couple of years later, Salvatore joined the company, and he is now leading all the activities around continued development of the open source.

Now, a word about Salvatore. This guy started Redis, and the first version, I believe, was out in ‘09. He did it initially as a part of the project he was doing for Telecom Italia. He was looking for a very fast database, he couldn’t find one, so he decided to develop one by himself. He didn’t have great expectations from that and just put it out as an open source.

That was ’09. Around 2011, Redis started to pick up and became extremely, extremely popular. Today, it’s one of the most popular databases in the world, and definitely one of the most popular open source projects in the world. So, that’s a little bit of history of Redis.

Customer Segments

Michael Schwartz: Most of our listeners probably know that Redis is used by a myriad of organizations, and Redis Lab has thousands of customers. But, when your potential market is so broad, how do you segment the customers?

Ofer Bengal: Redis is very horizontal in nature, so it covers many, many use cases for modern applications, and that’s why you find Redis today in almost any enterprise in the world, not to mention startup companies, developers, etc, etc.

We basically sell to all industry verticals, so just to give you some names.

In banking we sell to American Express, Wells Fargo, Credit Agricole.

Financial Services: MasterCard, Visa, Intuit, Fiserv.

eCommerce: Walmart, eBay, Starbucks. Social: Twitter, Twitch. Media: MSN, DreamWorks. Advertising: Havas, Rev Mobile, Outbrain. In technology: Apple, Cisco, Dell.

In communication, Comcast, AT&T, Verizon, Vodafone. And so on and so forth, in other industry verticals.

Michael Schwartz: I guess, my question was, because you can sell to anybody, do you break up the customers into different types and segment them in any way, when you look at them from a marketing perspective?

Ofer Bengal: We look at few dimensions.

First of all, we look at what are the capabilities of the database.

We map that into use cases of modern application. And then we map that into industries. We have all these matrices of mapping, and we have white papers and explanations for each of those. The idea is that, today, the world of databases has changed a lot from like 10 years before.

Originally, you had one database type that fits all. So, if, I don’t know, Citibank decided to use Oracle, they’ll adopt Oracle for everything. This has changed a lot because today it became more silo business.

Certain databases are good for certain type of use cases, and these use cases are very typical for certain industries. So, today, you find in a single Enterprise multiple databases not just that, with the single application, even a simple iPhone application, you can find five, six, seven different databases serving it.

Everything is changed, and that’s very good for us, because the market is a way more diverse now, and there is room for new entrance, and so on and so forth.

Value Prop

Michael Schwartz: What’s the value proposition of the Redis Labs’ commercial offering?

Ofer Bengal: What we did with Redis Enterprise is, we took it to the next level in terms of being suitable for Enterprise use.

There is a difference between an open source project, which can be very powerful but misses several elements, which every Enterprise requires.

In terms of scalability, we offer infinite scalability, various degrees of high availability with instant auto failover, better performance, more stable performance, and tons of other things, which make the product more robust and suitable for Enterprise use.

Revenue Streams

Michael Schwartz: What are the revenue streams for Redis Labs?

Ofer Bengal: Monetizing open source has always been a challenge, as you may know.

It all started with Red Hat providing support for the software. We never thought of this as a good business model, so from day one, we decided that we do not want to provide support to the open source. Until this very day, we do not.

Basically, the idea was, “let’s add value to the commercial software and sell it.” And that of course comes with support.

Today, we offer it in various delivery models. The first one is what we call database-as-a-service.

Database-as-a-service means that it’s a fully managed service, fully automated service, whereby the customer signs up in a zero-touch manner, and everything is done automatically by the system without human intervention.

He signs up, they create a subscription, and then they start to use the database with their application, without us being involved in the process. And they pay starting with on-demand, which means, as much space they consume in terms of gigabytes, terabytes, petabytes, etc.

And then we try to convert the large customers, the large on demand customers to annual subscription, where they commit for a year. That’s one part of the business. And then, the other part is the more conventional software selling part, whereby we sell annual subscription to our software.

Again, we quantify by what we call shards. Shard is a process of a database that has certain capabilities in terms of throughputs and amount of data that you can process, etc. But, anyway, all-in-all, we quantify what we sell according to the size of the database in general, and then you can get it, either as a fully managed service, or as a software that you install by yourself and operate by yourself.

Breakdown Of Revenues

Michael Schwartz: Can you share with us any percentages of like what percentage is database-as-a-service or versus software?

Ofer Bengal: When the company started, we only offered database-as-a-service, we didn’t have fans at the beginning to create, build a large field operation. So, we decided to go for the zero-touch model, which is not very costly. We started there.

A few years back, that accounted for like 80% of ourselves. Today, it’s changed. There we have a significant sales force. Today, we have around 60 to 65% of our business is driven by field operations, direct sales, and only the rest is database-as-a-service.

R&D: Open Source V. Commercial

Michael Schwartz: With regard to R&D, how do you prioritize how much you should invest in the open source core part of Redis, and how much to invest in the commercial offering?

Ofer Bengal: With the open source, there is Salvatore, with a group of developers. He still leads the development of the open source. And then, we have a way larger team, working on what we call Redis Enterprise, which is our commercial offering.

The way to decide what to put here and what to put there is kind of very sensitive. Basically, our decision was that this fine line is drawn, so whatever it has to do with new functionalities, in terms of commands, data types of the database – that will go to the open source.

When it has to do with the deployment of the database in a large-scale, and all the ongoing operations of the database – that will go to Redis Enterprise. I think it works well for us because the database is continuing to be very popular, whereas we are doing good with selling the commercial offering.

Licensing

Michael Schwartz: Redis is the poster child for what some call open source strip mining. Amazon, Google, Microsoft offer hosted Redis platform, and despite huge revenue, they haven’t contributed a fair amount to fund innovation of the platform.

If I understand the situation correctly, Redis changed some previously BSD-licensed software to a non open source approved license to prevent this opportunistic behavior.

My first question is, why not just change to the AGPL?

Ofer Bengal: Although AGPL provides a little better protection than BSD, because you need to contribute back basically any enhancement that you do to the software, we didn’t feel that this is a showstopper for a cloud providers to adopt a successful open source projects, and I’ll explain.

What happened in recent years is, as the cloud business grew, cloud providers such as AWS are adopting any successful open source project, making it a fully managed service and selling it for tens of millions, in the case of Redis, hundreds of millions of dollars per year.

Now, they did not contribute anything to the development of those open source projects, so it becomes some sort of unfair competition because they use their monopoly power, and the fact that people come to those clouds for the infrastructure, and eventually, they see all those data services that AWS offer.

To me, it’s exactly the same situation as the antitrust thing that we had in the past with Windows and Explorer, if you recall that. Microsoft was basically giving you Explorer for free, and by this, trying to capture the browser business. And there’s nothing we can do about that at the moment.

What we thought was that we didn’t want to change the license of open source Redis all together. First of all, you can’t change history, and secondly, we didn’t think it’s good, in terms of adoption perspective, etc.

So, what we decided to do was to leave the core of the open source BSD as it was. But certain new enhancements that we are adding over time, will be under a different type of license, which is still source available.

So anyone can take the source code, change it, do whatever they want with it, except one thing – they cannot sell it as a product. Because we think it’s not fair.

So, this is what we’ve done, the initial concept was called Commons Clause, it was done together with a few other companies. We are now looking at ways to make it even better for various reasons. I will not get into that today, but the concept is the same.

We think that we need to protect our rights as developers of certain software products, and avoid cloud providers for just taking that, without investing anything and monetizing it big time.

Community Reaction To License Change

Michael Schwartz: Redis Labs has Enterprise software that they release, there’s this Enterprise platform, and there’s Redis Core – do you think that the open source community is overreacting here? Isn’t this a common practice?

Ofer Bengal: With all due respect to the open source institute, I think that they are a bit detached from today’s reality, and for them, open source is like a religion.

I think that the environment has changed a lot since the ‘70s, or ‘80s, when open source started, with the emergence of cloud providers and what they do with open source.

I think that the market has its own dynamics, and that’s why you see all these different initiatives, not just by us, but other companies have started similar initiatives of trying to bypass the open source with something which is a little bit better.

AGPL-MongoDB

Michael Schwartz: In fact, MongoDB, who’s had AGPL in the first place, for some reason, felt that AGPL wasn’t providing enough protection.

Ofer Bengal: As I said before, we always thought that AGPL does not provide protection, because the proof of that, by the way, is that recently AWS announced what they call DocumentDB, which is basically MongoDB product. I guess they took some of what Mongo had under AGPL in the past, before they changed to SPPL.

And, practically, if you read the AGPL license, it does not prevent companies like AWS from taking the product and selling it.

Progress In Harmonization

Michael Schwartz: In the previous Open Source Underdogs podcast, Mark Shuttleworth, Founder of Ubuntu and Canonical, implied that both sides need to work together, that it’s normal for this type of evolution of license and business relationships to evolve.

But do you see any progress that this may actually be happening?

Ofer Bengal: Well, I hope so.

To be honest, I have a company to run, and we need to grow the business, so this is not our main goal in life. We will be happy to participate in any such effort obviously. I’m not sure that we have any interest to lead, any efforts like this, because we have other things in mind, but I definitely think that this is required.

How To Compete With Amazon

Michael Schwartz: Redis Labs has its own cloud-hosted database-as-a-service, and although Redis Labs is pretty sizable, no one has economies of scale like Amazon, Google, and Microsoft. How do you compete in that hosted market?

Ofer Bengal: The only way to compete is with technology, and what we are doing now is taking Redis to the next level.

Redis started as a cashing system. Later on, it became a data structure engine, and later on became a database.

What we’re doing now is taking it to the next level, which means, making Redis what is called a multi-model database, whereby you can model the data in many different ways, starting with key value, data structures, graph, time series, streaming data, search functionalities, AI functionalities, and some more.

The idea is that with that, which is, by the way, the new trend in the database technology, we will be able to cover many more use cases, and by that, capture a larger market share, so that’s our strategy.

All these new enhancements are licensed differently, with this new different license, which prevents the likes of AWS to adopt it.

So, that’s our strategy in terms of competing with these companies, because, obviously, in terms of visibility, etc, you can’t compare.

Sales and Marketing

Michael Schwartz: You mentioned previously that you have built up the sales force overtime. Can you talk a little bit about sales and marketing? I would imagine you originally started with a lot of inbound, but how have sales and marketing sort of evolved since you got started?

Ofer Bengal: As I said, in the first years, everything was inbound.

We started zero-touch, which means, we drive people to our website, where they can sign up, create subscriptions, etc. Those were the first years.

After we completed building the software product, this is very hard to sell online, just by putting it on the website, so we started to build field operations – that was around 2015.

Today, it’s pretty sizable organization, we have people on the ground in all parts of the US. We have a team in India, we have teams across Europe, we have a team in Israel. And when I say a team, they comprise sales rep, they comprise solution architect, so any sale is very technical in our case.

It’s not just a sales person knocking on the door and saying, “hey, I’d like to sell you a refrigerator or something.” You need to work with the tech people of the customer, analyzing their environment, their architecture, etc, and finding the best fit. So, this is done by our solution architects, and they work in teams with the reps.

Then, we have all the marketing organization, which basically drives demand, so they work with all advertising, with a CEO, we do tons of events. This year, we’re going to do 90 different events across the world – all of that drives demand, in terms of leads.

Then, we nurture the leads to become what we call MQLs, Marketing Qualified Leads. And then we have a team of what we call SDR, Sales Development Reps, which basically pick up the phone, call these MQLs, and try to see if there is substance there, if there is a project going on that we can tie in, etc. If so, they convert them to become what we call QSLs, Qualified Sales Lead, and those are handed over to the salespeople, to start doing the actual sales work. That’s the way we work.

We also rely heavily on partnerships, we have a few very strong partners, and those partnerships are very intimate. We really go to market together, both on the marketing side, and also sales teams work together. We work like this with Pivotal, with Red Hat, now with Microsoft, and some others. So, that’s very powerful for us.

Partnerships

Michael Schwartz: Can you just maybe elaborate a little bit more on the partnerships? You have sort of technology partners, like you mentioned Pivotal, Microsoft, etc. What about integration partners, or other types of partners – how do you work with those?

Ofer Bengal: We work with some of the large system integrators in the world.

I can’t say that so far we found the sweet spot of how to do that, in terms of, it seems that when you get to the larger and larger companies, and as I mentioned before, we sold and worked with the largest companies in the world. We have approximately 40% of Fortune 100 companies as customers, you need really to be in direct contact with their technical teams.

So, it’s very hard to do it second-hand, that you educate someone, and they do it. So far, we weren’t too successful with that, although we very much like to leverage that, obviously because that gives you much larger coverage, but so far, our successes in partnerships were more with IT companies, who basically try to sell the entire IT suite to those large Enterprises, companies like Pivotal, Red Hat, etc.

Challenges of Being an Israeli Company

Michael Schwartz: Has being an Israeli company presented any special challenges?

Ofer Bengal: The major challenge is me spending half of my life on planes, that’s the thing. Because I split my time between the Mountain View, where we have our headquarters, and Tel Aviv, which is R&D Center.

The flight time direct between Tel Aviv and SFO is 13 and a half hours, direct. And if it’s not direct, it can easily take like 20 hours or so. Anyway, I spend a lot of my time on planes and fighting jet lag fiercely, but that’s what I have to do.

Michael Schwartz: So, you get time to listen to podcasts.

Ofer Bengal: Exactly.

Redis In 20 Years?

Michael Schwartz: Where do you see Redis Labs in 20 years?

Ofer Bengal: I think that we have a very unique opportunity, which, in my eyes, and I’ve been around, is a once-in-a-lifetime opportunity.

We have a very rare combination of a huge market, which is being disrupted, and open source, which is extremely popular, strong value proposition, with the commercial product on top of the open source. And then, apart from that, Redis is the fastest database in the world. I don’t know if I mentioned that before.

I don’t know if you remember the first time you did a Google search, to me, it was like magic: You clicked and you got it immediately. Now, today, this type of performance is expected from any modern application, especially with Millennials, Generation Z, etc, they don’t have the patience to wait, they would like to click and get it.

Now, when you look at what happens under the hood, when you click, a request goes over the internet to a remote server, a database is involved, it has to get back to you, etc. If all that happens in more than 100 ms, which is 0.1 second, you start to feel uneasy, as if something is not operating properly.

Now, the database was always the bottleneck when it comes to end-to-end application performance. And Redis is the only database that can process tens of millions of transactions per second, and all that, it’s sub-ms latency.

And with that, we see ourselves today as the enablers for modern applications to provide what we call instant experience. By the way, this was always the selling point for Redis, and I’ll claim to fame, and we push very hard on it.

Today, we offer graph functionality, much faster than any other graph database, we offer document functionality way, way faster than any document database, time series, the same, streaming, we do so much faster than Kafka, etc.

So, whatever we do, we do way faster than anyone else, and this is basically our main setting point. So, today, people in the industry know that whenever you need something fast, there is only one solution, and that is Redis.

Going back to your initial question about where do I see Redis Labs in 20 years, I’ll start from the beginning.

I think that we have a very unique opportunity, which in my eyes, and I’ve been around, is a once-in-a-lifetime opportunity, and this is because we have the very rare combination of a very large market.

The database market is going to be $60 billion market this year, which is being disrupted for the first time in its history with the entrance of new type of databases, which some people call NoSQL, etc, and many other changes in this market.

Secondly, we are based on a very, very popular open source project, which is available and being used in almost any organization in the world. That’s very unique.

On top of that, we have a commercial product, which is robust and stable now, which provides many value adds to large Enterprises on top of the open source.

Add to that the growing demand for instant experience, today, people would like everything to be instant, we provide that capability.

And add to that the fact that Redis runs originally on RAM. RAM is relatively more expensive than other media such as SSD, or obviously disk. So, in the early days, it was like a little bit of a hindering factor for wider adoption of Redis because of the costs associated with the underlined resources. But, today, there is another trend, which works for us, and this is the trend of new memory technologies, persistent memory technologies by companies like Intel, Samsung, etc.

If you compile all this together, it seems that we are on the verge of a huge opportunity, and
it’s basically for us to do it, or not to do it. And this is what I always tell the people in the company, “guys, it’s in your hands.

If we don’t make it a very big multibillion-dollar company, this means we did something wrong.” That’s it.

Advice For Startups

Michael Schwartz: The goal of this podcast is to help entrepreneurs who are starting businesses around open source. Do you have any advice for entrepreneurs who are just getting started about how to use open source software as part of the business model?

Ofer Bengal: Do not neglect the business model. Think about your business model from day one. Because when it comes to open source, that’s the most challenging issue. You know, you can build a great product, great technology, which everyone likes, you want to make a penny out of it, so think what is your strategy, in terms of what do I sell, how does this compare with what’s available in the open source for free, and basically how do I make money.

Credits

Michael Schwartz: Ofer, thank you so much for sharing your wisdom.

Ofer Bengal: You’re welcome.

Michael Schwartz: Transcription and episode audio can be found on opensourceunderdogs.com.

Music from Broke For Free and Chris Zabriskie. Our audio editor is Ines Cetenji.

Production assistance and transcription by Natalie Lowe. Operational support from William Lowe.

Follow us on Twitter, our handle is @fosspodcast.

Tune in next week for interview with Ravi Mayuram, CTO of Couchbase.

Until then, thanks for listening.

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

Introduction

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.

Revenues

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.

Partnerships

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.

Funding

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.