MapR’s Top Execs Sound Off on Hadoop, IoT, and Big Data Use Cases
When it comes to putting big data into action, there are few companies doing what MapR Technologies is doing. From implementing massive Hadoop clusters on Wall Street to preparing for the next wave of distributed computing with the Internet of Things, the company is a leader in its space.
Recently the company’s top brass, including founder and CEO John Schroeder and Matt Mills, who was recently hired as MapR’s president and COO after two decades at Oracle, sat down for a telephone call with Datanami. Here’s part one of the highlights of that interview.
Datanami: Matt congratulations on the new job at MapR. Please tell us a little about what you did at Oracle, and how it prepared you for this job.
Matt Mills: I started at Oracle in ’93 as a sales rep. Over the course of 21 years, I was fortunate to be in the right place at the right time. I ended my time there last year. I ran the North American field organization as a GM–all the hardware, databases, middleware, a good chunk of the applications, all the converged architecture. We had about $4.5-billion business and about 8,000 sales guys. I didn’t intend on staying that long to be honest. It’s just one of those things. I took six to seven months off and I started looking around and was fortunate enough to come across John and MapR.
Datanami: Does MapR feel more nimble and more able to quickly adapt than Oracle?
Mills: It might not seem to the naked eye but Oracle still, even today, a very entrepreneurial company. Direction and strategy and brand and the constraints are set at the top, but then it’s up to you as the business owner to go get it done. But it is refreshing. I was telling John, he’s done a good thing with the company in terms of everybody’s a shareholder, everybody’s vested. You walk into an opportunity like that, it’s pretty energizing actually. I’ve been pretty stoked from day one. The only criticism I would ever have is that we need to get out and make people aware of what is, because it’s terrific.
Datanami: If that’s MapR’s only problem, that’s great!
Mills: We’ve got challenges. But they’re good challenges. If there weren’t things to go do, they certainly wouldn’t have hired somebody like me. There’s things I can come in and help with. I’m trying to stay out of the way of the things they’re doing well and try to put forth a little bit of elbow grease on the things that I can help and add value in. But I’m pretty picky. I passed on a number of things, just because I was trying to find the right thing. I feel very fortunate to have stumbled across John.
Datanami: What skills and capabilities will you bring to bear at MapR?
Mills: If you spend time at Oracle, as I did, and you had some success, you can’t help but walk out of it with some degree of operational excellence. So I’m going to bring some energy. That’s one of the things I have, is energy and passion. And over the years, if you do things right, you build a pretty good network of people that would like to go do it again. So I think those are some of the things that I’ll bring to the team.
Datanami: Hadoop is evolving at a tremendous pace. Is that constant change helping or hurting?
Mills: It slows my down my ability to absorb it all. I’ve attached myself at the hip to John because this is second nature to him. He can go as broad or as deep to anybody in this space. I try to soak up as much as I can from him. But you’re right. Therese’s a lot of stuff. But I’m trying to be a quick learn.
Datanami: John, what impact is the constant evolution in the Hadoop space having on your customers?
John Schroeder: We’re seeing the market continue to take off like crazy. It’s becoming less Hadoop centric and more about big data use cases. That’s a good thing. We’ve done really well with the early adopters and innovators in the market. But getting into the mainstream, it’s about big data use case. As Spark emerges as a great new compute technology, and Mesos, YARN, Myriad, more interactive SQL, graph databases. We’re also getting into messaging for IoT where you distribute a lot of the processing to be able to execute in a multi-tier environment for IoT.
This is all good for the marketplace, and it creates applications that couldn’t be created before. We’re the engine for a number of ad tech companies. They couldn’t do billions of auctions a day without this new technology and the innovations there.
If you look at the traditional enterprise market, we’re helping healthcare providers make the right treatment recommendations. We’re revolutionizing how the auto industry works, where basically you’re going to pay premiums based on how much you drive and how you drive, instead of going with the less accurate demographics and historical driving.
These are all applications that you couldn’t build before. So I think the market is moving to a space where the conversation is more about apps and use cases. And the feature set, both what MapR brings to the table with our propriety engine and what we do in the open source and what others contribute to the open source as well–just gives the customer a broad range of technology to choose from.
Now, we have to be prescriptive and consultative to our customers so they can talk to us about their customer about use case. We can help them say, hey Spark is the best technology for that, or Apache Drill is the best technology for that, or this should be in a graph database like GraphX or Titan. So the superset set of capabilities is a good thing to drive business use cases.
Datanami: There’s talk of HDFS being replace for some things with Kudu or Mesos replacing YARN, just like Spark is replacing MapReduce. Will we still be talking about Hadoop in five years?
Schroeder: Probably not. You’re still going to have a need for a big data engine. But you’re right–it’s not just Hadoop. It was really sleepy in the Hadoop space until two years ago. But now you’re seeing Spark, Drill, Impala being major innovations. There’s some integration work being done in a project called Myriad, which combines the distribution and services of Mesos with the resource management of YARN.
But these are all just new innovations that add value to the platform. I don’t think people are left behind if they choose the right platform. If they put their data into MapR and were doing that for MapReduce, well they’re going to be able to process that with Drill for interactive SQL and they’ll be able to process it with the HBase API for NoSQL processing. They’ll be able to do messaging at scale with that same data. They can run YARN managed apps against it.
Our strategy is to enable all those compute and resource management platforms on top of the same data so you don’t have to implement multiple clusters and create multiple silos of data. So that’s why we’re particularly quick to embrace YARN, Mesos, Impala, and Drill. Those are all good things for our customerss and good things for MapR.
Datanami: As you go into a sales situation, do you spend a lot of time educating customers on what’s even possible with today’s big data technology? How to bridge gap between what customer needs are and what the capabilities are?
Schroeder: That’s a big part of our job, to go in and in some cases respond to what the customer use cases. But frequently we say, hey these are the types of things that companies like yours in your segment can do with big data, and make the possibilities visible to them.
Then there is a matching of the broad range of compute technologies with your use case. Is this real time streaming, is this interactive query, is this dashboarding, is this an app that needs to be distribute across multiple data centers or perhaps do real-time messaging?
Stay tuned for part two of our interview.