In San Jose (CA), we meet CEO and Co-Founder of MapR, John Schroeder. John talks about his story how he came up with the idea and founded MapR, how the current business model works, as well as he provides some advice for young entrepreneurs.


Martin: Hey, have you heard of big data, Hadoop and all those kind of stuff? Maybe then you have heard about the terminology of MapReduce as well. Today we are in San Jose at MapR. Hi John. Who are you and what do you do?

John: Well, I’m John Schroeder. I’m the CEO and co-founder of MapR. I started the company a little over six years ago and named the company MapR, named after kind of the seminal algorithm for big data. So Google fellows wrote a paper on MapReduce in 2004, that really started the whole inspiration behind Hadoop so we named the company MapR. We’ve been off to the races and it’s a wonderful, wonderful opportunity. We are having a lot of fun running the company.

Martin: Awesome. What did you do before you became an entrepreneur?

John: Wow. I’ve been in startups for quite a long time. I was a general manager at a public company back in the nineties. I spent over twenty-five years in database, storage, big data, business intelligence. So this is really my fourth start-up. First one I founded but my four start-ups.

So I was at Brio Technologies in the nineties. And Brio got out in public and listed on NASDAC in 1998. So there was an exciting ride back in the nineties. Then I was CEO of a company called Rainfinity which had a file virtualization switch that was acquired by EMC; became a very successful product line for them. Prior to starting MapR I was CEO of a company called Calista Technologies. And there we actually wrote software that ran on a GPU that would virtualize the GPU and render and remote 3D graphics and multimedia. That was acquired by Microsoft and became the Microsoft RemoteFX, remote display protocol. So it’s been database, storage, enterprise software.

What was obvious for me in 2008 were that macro trends on big data. Companies just needed new ways to connect with their customers and ways it would provide value to the customers. Health care providers need a more accurate way to prescribe treatments for their patients. Wide range of governance across financial services and telecom, forcing them into Big Data solutions for storing email archives and call data record archives and telecom providers and carriers. So those macro trends were obvious and then you could see this new wave technology around things Hadoop. So that excited me to stay in that enterprise software space and go give value to those customers with a big data solution that would really serve these macro trends.

Martin: And how did you then start with MapR? I mean especially did you talk to some potential clients before or did you talk to investors or maybe just friends, validating your idea?

John: Yes. All through 2008, I built a really good rolodex of CIO CTO’s across industries, across geographic territories. And I really just start out with open-ended questioning. Like, what are your big challenges for the next five to ten years? Why are they challenges? What happens if you are able to accomplish these challenges? What happens if you don’t? And through that that really formed a lot of the basis for the big data was important. This was in the top two or three priorities for almost every individual I talked to across the industry.

Then I got more into, well what sort of technologies are you trying to use? And what do you like about it and what don’t you like about it, and how ideally would you like them to work? The macro trend for big data was obvious. Which technologies, you can imagine 2008 it wasn’t just Hadoop, it was Hadoop and Cassandra, MongoDB, and CouchDB and Volt DB. I mean there were just so many emerging technologies that the signal-to-noise ratios there wasn’t quite as strong but you could see a little more market share for Hadoop.

But more importantly, my co-founder M.C. Srivas was working at Google at the time. He and I can look at and see how can we grow Hadoop to really cover all the big data needs? So all these technologies started out in kind of a niche and Hadoop’s niche was batch predictive analytics and scale. Well that’s a part of what the customers need but they also need interactive, they need primary storage, they need real-time, they need messaging. So one of the reasons we chose Hadoop was being able to see that we could grow the technology; really handle one hundred percent of the customer’s use case.

So then based on that, well in the valley for a long time you know you’ve got Sand Hill Road and you find your friends on the Sand Hill Road and we put together a good business plan. In my case I likes to have a consortium of two really tier one investors at my A-round. So in this case we chose Lightspeed Venture Partners and NEA and they split their round. And then if you’ve got two investors with deep pockets at the table. If you’ve got good investors they’ve got great networks to talent, they’ve got great networks to customers. So I met my senior vice president of product management through NEA and I actual met my co-founder through Lightspeed. So we put two tier one’s in that A-round and that was very important and that sets up for future funding because when you get around to your B-round, well to get another tier one investor, you need to have tier one’s in you’re A-round. In most cases.

Martin: So this means first you validated your idea with some of your connections, so to speak. Then you used your connections on Sand Hill Road for raising some money. What was the next step? Did you fully build your kind of platform and acquired tons customers already? Or did you only ship the MVP and try to validate whether there is some kind of customer demand there?

John: It took us about a year and a half to get into beta, so we had this pleasant year and a half experience. If you can imagine, during the company you set your own milestones and then you’re the one who judges whether you made the milestones or not. So it’s kind of the least pressure stage of the company compared to now where we’ve got a quarterly number that me got to match to every quarter.

What we did is we kept in contact with those forty some odd customers we had done the primary research with and then we grew from there. We kept adding more customers to continually validate the concept and then put prototypes in front of them and get their feedback. So once we got to our beta period I think we had thirty-seven companies in beta and we exited our beta program with just under a million dollars in sales. So by staying really in contact with those customers we’re building the product they needed so it’s no surprise that they brought once the product was ready to run.


Martin: Let’s talk about the business model of MapR. So John, what are your target customers?

John: We’re a platform sales. So it’s not a Jeffrey Moore crossing the chasm find a little niche. We sell to just about everybody in the top financial services market. Telecoms are our number two market segment. We do about twenty-five percent of our business to web 2.0. So companies like comScore, Rubicon Project, Millennial Media, companies like that. So it’s very horizontal, we’ve got customers who bought over a million dollars worth of software in eight different vertical markets. We’re about seventy percent domestic and about thirty percent rest of world. The uptake for the product’s been really strong in Japan, Korea as well as other countries you’d expect in EMEA as well.

Martin: How is your product or product offering comparing to competitors offering?

John: It’s kind of a continuation of your question on business model, which is, like everyone in Hadoop space or in a big data space the way to build ubiquity through a platform is through open source. I mean you get tremendous innovation and you have comfort from the customers that they’re using something that’s industry standard. There is basically a reference implementation available and in our case and provided by Apache software. So that builds ubiquity, that builds the polls in the market, that gets customers very comfortable. Then what we did is we looked at Hadoop and its very early in its life technology lifestyle. So it’s open for massive innovations. You know, how did you take this badge predictive database and really make it really interactive in real-time and then even support real-time messaging. So that’s where we built our differentiating technology as a platform that can run that open source.

And so that’s the concept of how we ship our product and that drives a different business model. So rather than just selling services and support around free software we’ve got value in our software. And we saw the value that software customers in the form of software subscriptions. So we end up being unlike most of the others in the space. Very high gross margin, we are less capital intensive which for an entrepreneur, every dollar you raise is also taking some stock out of your pocket. So you have to fully capitalize your company and make sure you can spend at the rates to be successful and build the company that you want.

On the other hand, you don’t have to create a business model that’s too capital hungry. And that’s very, very negatively impactful for a lot of companies who tried to build open source companies because they’re so capital intensive that they end up raising hundreds of millions of dollars and then having a hard time making those investors happy and then also maintaining the equity values for the employers of the company as well.

Martin: Big data is a buzz word nowadays. And lots of people know that, for example, you take some data sources, plug it into or push is into the HDFS, and then you have some kind of batch analytics processing. You also said that you have this kind of real-time analytics solutions. How does that work?

John: If you really look at end to end use case that the customers want it never going to end at batch. So I’ll give you an example of Rubicon project, runs on of the largest ad exchanges today. So if you look at what they’re doing is they’re placing online advertisement as people are browsing the web. So there is a primary storage component of that, they have to store all the auctions, all the bids, all the asks, and the outcomes, right? And then there’s a new analytics piece in it which is they need to analyze those auctions and come up with yield estimates for certain types of page views and certain types of publishers. But then in real-time they have to mediate between thousands of brands and publishers, right? And so it’s a real-time activity. Well they can do all that in MapR. And if they tried to use an alternate to MapR they’d have deploy multiple technologies, multiple silos of data and deal with the complexity of data governance across platforms. So of course they prefer to do that in MapR or we could do a hundred percent of the use case on one platform.

Martin: When you think about your customer relationships and especially the question relates to, how do you manage them and nourish the customer relationships you are currently having?

John: You have to make that the number one priority for your company. I mean, your customers are not your customers, they’re your partners. I mean you have to do anything possible to make sure they’re successfully with the use cases they’re deploying on your technology. So that’s key and if you do it properly you’ll have different conversations with them. Sometimes, they’ll be coming to you and saying, “Hey you know, we really love your products but we really need this additional feature set and.” And you’ll learn a lot from them. And other times when you’ve got brilliant engineers and a brilliant CTO, they’ll be bringing ideas to the customers and saying, “Hey, what if you had this? How would you implement this as part of your next use case?” And you’ll be enlightening them as well.

So that partnership is very, very important. We’ve had a wonderful customer advisory board. I mean another successful program we’ve had here is to take our top twenty/top twenty-five customers out of their offices, get them in a good environment, share with them what our roadmap is and then just listen to them as far as what are your use cases? What are your challenges? What do you want us to do more of? Those are just fantastic events as far as really helping us shape our roadmap.

Martin: How often do you do these events with these twenty-five customers?

John: The formal events are once a year. We’re connected to the customers constantly but once a year we try to pull together really representatives cross industry and across geo and get them together in a room, and it’s a wonderful experience.

And I would say three years ago they took something like priority number twenty on our list and said, “No, you don’t understand. That has to be in the top five.” And so we moved it up and it was a great advice. Last year M.C. Srivas my CTO and co-founder, he gave them a talk on how to do scalable messaging on MapR. And that’s where you can see them learning and challenging him, “Well why wouldn’t you do it this way? Why wouldn’t you do it that way?” And he’d say, “Well, have you thought through the aspects of high availability or performance for things like that?” So it’s a great two-way conversation and it’s, you know, if you can pull together a good set of customers likes that they’ll really push you in the right direction.

Martin: John, this is your fourth start-up and the first one that you started yourself: over those four start-ups, what have been the major obstacles that you’ve seen and how did you overcome them or manage them?

John: I mean I’ve been very fortunate. I have had one public company and two successful acquisitions and then MapR’s been a really fun ride. And to be honest with you MapR’s straightest line from A to B of all of them. The other ones all needed some sort of strategic change. So with MapR we have been able to set a course and stay very, very close to that course over time.

The challenges, I think you have to be resilient if you think about the different attributes especially in Silicon Valley or in tech that people value. They value of the brightest of the bright. And then work ethic, the work ethic in tech is epic. People just work all their waking hours and they love it like. It’s not their grueling you know, unhappy working at ten o’clock at night, they’re enjoying it. They enjoy the challenge and they enjoy the technology. So you could say the brightest of the bright, the hardest working, teamwork’s a given, things like that. But to be honest with you, resilience is probably the most important attribute because you know, you think about six years ago Srivas and I were talking and saying, “Well, in a couple years we’re going to have our software deployed running critical risk and fraud algorithms at the largest credit card company in the world. How easy it that going be?” Right? So you can image the technology challenges and things like that.

So you have to be resilient and as long as you go after a great a market and you team with really, really talented people, you’ll be able to go through those ups and downs. And overall hopefully the trend is up.


Martin: John, imagine your child is coming to you and says, “Daddy, I would like to start a company and this is my idea.” What general advice would you provide to her?

John: Well for children or friends or other potential entrepreneurs, I think the first one is to look at the market opportunity. What’s the macro trend that’s going to drive your company going forward? So many of us, like I’m a computer science grad and a software engineer at heart. So we get so excited about technologies and I don’t know about you but like, even when I was a kid, I didn’t know how everything worked. I took things apart whether they were engines or electrical devices or whatever. And you get really, really interested in technology but you really have to step away from that and look at what’s the trend that’s going drive this? How are you going to find a very large addressable market?

Because, there is kind of three levels of risk and a technology start up.

  • The highest order risk is the big market risks. I mean, if we all wake up tomorrow and big data is no longer important, I’m in trouble. It’s like how do you reposition this company and technology you built.
  • The second level if your technology company is your technology. Does the product work as advertised? And you might be able to fix a problem there. You can say, “Hey, the products needs some work. Let’s raise a little more money and give engineering another year.” So that’s a little bit fixable.
  • And then there’s really kind of an executions risk. Which is you’ve got a great market, you got a great product, then you know you got it to bring market properly, you have to service your company properly. And that’s the easiest to fix, right? I mean you could afford a few mistakes there and you could resolve things pretty well.

So if you look at those three elements of risks, before you start the company make sure the market’s there. Make sure there’s a huge addressable market you can go after and then you really set yourself up for a great ride. And you should be able to build a great product and you should be able to build a team around it that can execute well as well.

Martin: One question regarding your assessment of the future. So imagine we have 2030 or 2040. What is your perspective on how companies will use data and build their database and data pipelines?

John: The next big wave is internet of things. So we’ve gone through this whole wave of just ubiquity of computer connectivity and storage and it’s open up a huge new opportunity that we’re addressing today. But now you’re starting to see machines talking to machines and it’s going to a be a whole other wave of big data.

So you’re going to see it move from being a central repository if you look at when people talk about MapR and Hadoop they’ll say, “Well, I’ve got x amount of data in MapR”. When you get to Internet of Things you’re going to have to distribute that workload again. So if you look at some of the technology we’re building now it’s to have a small form factor that let’s say on a smartphone, a little bit larger form factor on an edge device and then still have the cloud deployment that could be very scalable but you’ll see it move from a centralized model to a more decentralized model.

Martin: The very old model was one central database. So now we have in 2.0 which is the big data, Hadoop where you say, “Okay. At least we have some distributed files and distributed calculations and so on and now the thought wave would be: Every device, whatever it is we’ll do some calculations for itself.

John: Exactly. And it’s really not new. What we do as an industry is we distribute workloads and then we consolidate workloads and we distribute workloads. So IOT by definition, if you have; whether their devices, servers, automobiles, smart phones, you’ve got a broad number of technologies out there that need to be able to communicate, make decisions locally and then just forward on what’s required to the next layer of the stack. So I think you’re going to see the last few years has been how do you scale a large cluster for running something like Hadoop or you’re going to see more decentralized processing with local decisions being made at the end point, at the edge, and then back enough more a central cloud.

Martin: John, what is the challenge or what keeps you up at night and saying, “Well I need to fix this?” Or what is the one thing you are thinking of?”

John: Probably the toughest thing of being an entrepreneur you can think of the initiatives faster than anybody can do them. I mean, I can look back on different documents that we wrote back in 2008 and we still haven’t implemented some of that. In the meantime we’ve come up with a thousand more ideas. So there’s just an endless amount of value you could bring to the market and how you can do that in a way that you can bring to market, package it so the market can even understand it. I mean the rate of innovations if very higher. So while all engineering organizations worldwide can’t get done everything they want to, it’s also the market that had to be able to absorb it fast enough as well. So I think that’s the challenge, you got the reins on the team of horses and you think, “Wow, I can make them go faster,” you have to go at a speed that the market can really absorb the technology.

Martin: Great. Thank you so much for your time and for sharing your knowledge.

John: Right. Thanks for having me on the video today.

Martin: Great, John. So next time you are thinking about data and big data, check out MapR. Great. Thanks.

John: Thank you.

Martin: Thank you so much.

Comments are closed.