It’s not every day you get a former NASA AI specialist to stop by for a beer and to chat big data. But sometimes the stars align. Monte Zweben, co-founder and CEO of Splice Machine, shared his favorite California-brewed IPA with me, Racer X, and kicked off with why the company was started.
Andrew: All right, I’m Andrew Brust with Datameer, and I want to welcome everyone to this episode of Big Data and Brews. Today we have with us Monte from Splice Machine. Monte, maybe you could tell us a little bit more about yourself and the company.
Monte: I’d be happy to. I’m Monte Zweben, I’m the CEO of Splice Machine. I’ve been doing software for quite some time. I started my career as a NASA AI scientist, and I’ve done a few startups including one called Blue Martini Software that went public and has been in the marketing space. Part of Splice Machine stems back to the troubles that customers had way back when with Blue Martini’s software, and being able to really scale their applications. The lion’s share of the problems ended up being really the database that was powering the application, so I sought to build the next generation of operational databases, and that’s really what Splice Machine’s all about.
Andrew: Blue Martini was a product in the e-commerce space, if I recall correctly.
Monte: That’s correct. Blue Martini was the first end-to-end e-business application suite, started with e-commerce transaction engine, and then it had a true campaign management application for omnichannel interactions all the way to in-store products at the time.
Andrew: We have a couple of heritage pieces in common. We’re both made of New Yorkers.
Andrew: I’m still there, though. You sold out and went west.
Monte: I did.
Andrew: But that’s okay. I’m in the minority.
Monte: I wore my Met hat.
Andrew: There we go. Not sure if that’s a good thing or a bad thing, given how they’re doing.
Monte: It’s a test of fate.
Andrew: You believe.
Andrew: Actually once upon a time I was with a consulting firm that ended up going against Blue Martini for a job, and you won, but you’re welcome here regardless.
Monte: Thank you very much.
Andrew: Maybe we can connect some dots later between your background, even pre-Blue Martini, the AI stuff at NASA and so forth, and what Splice Machine is doing today. Before we connect the dots, let’s get them both sketched out, I guess. Can you tell us about Splice Machine’s novel approach to database management? We tend to have categories in this field, and firm delineations between them. Sometimes those categories meet.
Monte: Yeah, thank you for that teeing up the classification of what we do. There is a lot of database technology out there, obviously, and first and foremost we have our traditional relational database management systems out there, the Oracle/IBMs/Microsofts of the world. Those systems have started to hit the wall from a performance perspective, they’re expensive. The database companies have addressed this in a very unique and I think positive way by providing scale up architectures that have highly engineered systems that perform in very unique ways and are incredibly powerful. There’s a cost associated with them.
Andrew: I was just going to say, they can be expensive, too.
Monte: Many zeroes.
Andrew: To the right.
Monte: To the right. The world moved laterally looking at some of these performance issues, and the NoSQL world emerged, and they developed an architecture of scale out. Scale out’s really powerful. It basically is the approach of instead of using one beefy machine, you use many inexpensive machines to operate on your data to store your data, and that distributed architecture gains you the ability to scale out. NoSQL was great for that. The problem with NoSQL is they threw the baby out with the bathwater, and now you had a database architecture that scaled, but wasn’t expressive enough to really build enterprise applications.
Our perspective on this was “Okay, how can we get the best of both worlds?” We are a relationship database management system that supports full SQL on a scale out architecture, and so we’re a scale out SQL database. There are a couple of different kinds of them out in the marketplace as well. One might be called a newSQL architecture, and these are the vendors who saw the same opportunity of scaling out SQL, but built the whole stack themselves. They had to build a distributed lele system, they had to build key value stores to store the data. They had to build all of the synchronization mechanisms necessary to do distributed operations. It’s a lot of work.
Then there is another category, which is SQL on Hadoop. We’re one of those. SQL on Hadoop is all about using the Hadoop stack as its underpinnings for a distributed file system for distributed computing, and all of the database technology we built is on top of the Hadoop stack. There were a few entrants to the SQL on Hadoop space, but the one really unique thing about Splice Machine is that we’re a transactional database, much like Oracle, or IBM, or Microsoft. We can power applications. We provide the asset semantics, which is the technical term for being able to handle concurrency, and we’re the only ones that can really do that. That’s our core differentiation. Net-net, the short answer to your question is we’re the only transactional SQL on Hadoop database.
Andrew: All right. I’ve got some follow ups for you on that, and I know you have stuff to tell us about the next generation of your product, so we’ve got more than enough to talk about. I want to take a sip of beer, but before I do that it’s traditional for you to tell me what I’m about to drink.
Monte: This is Racer X. I like IPAs, this is one of the better IPAs out there. It’s relatively local, and it’s a California brew, so I thought you might dig that. I only brought one bottle of it because we’re doing this midday.
Andrew: Yes, it’s business hours. Absolutely. Cheers.
Andrew: All right.