 James Phillips, co-founder of MemBase. Hi, I'm James A. Valante from Wikibon. You guys look like you're straight off of SportsCenter, I gotta say. That's the idea, we want to be the ESPN of tech. Yeah, absolutely, it's awesome. So I was reading your signs as you guys are related to Farmville? We are the database behind Farmville, so it's a... 250 million people, I mean they're throwing off some serious data. Does your database scale? Yeah, absolutely, I mean it's hard to imagine a better customer if you're trying to prove scalability. I mean those guys are... Yeah, you got any proof points? It's pretty impressive. So tell us a little bit about what's going on with MemBase. How's it that we want to drill into Zynga? Because Zynga is one of those companies that not a lot of people have heard of other than the games, but they're massive, they're massively growing, taking over San Francisco. I heard they're the largest, you know, employer in terms of space in San Francisco. Yeah, so... We want to get into that scale, but let's talk about MemBase first. Sure, so MemBase is a, it's a database, and it's a distributed database management system. In other words, it takes your data and smears it across a lot of servers. Instead of the traditional model where you've got a big box that runs your database and if you need to store more data and serve more transactions, you get a bigger box. With MemBase you just get more low cost boxes and it's designed to automatically spread your data around as you grow and shrink the cluster. And it is optimized for very low latency, very highly responsive access to that data, which is very important in an organization like Zynga, where you've got people who are playing games in real time. Right, if you want to buy a sheep, you want to buy a sheep right now. You don't want to wait five minutes for your sheep to show up on your farm. And frankly, Zynga wants to collect the revenue right now, right, before you leave. And so having a system that is optimized for being very, very responsive is good for the user, it's good for the business that's running the database. Are you guys all pure software? Are you applying as involved? 100% software. And so you don't really care solid state brand? So we are, we have a built in, wow, you're technical for a sports center guy. We have a... I've heard those words before. Yeah. Memcache, Membase, we've had Fusion I.O. on theCUBE. Okay, great, we're actually partnered with Fusion I.O. They help solve a problem. They absolutely do. They start to talk about low latency and locking becomes an issue. Absolutely. Just to scale out. So Membase is a, it's got a multi-tier storage management model. So what happens is, we've actually got a built in Memcache D layer. So the guys that wrote Memcache D wrote Membase as well. And so there's shared, you know, there's a lot of synergy between those two projects. And so Memcache D is actually built into Membase. And as you put data into Membase, it ends up sitting in RAM initially and then it bubbles its way down to an SSD layer. Exactly. And so things bubble up and down the latency cost stack in an automated fashion. So you might use a Fusion I.O. as a server class memory that's persistent. Yes. Another layer in the hierarchy. Exactly. Which supports that. I mean, you're throwing multi-cores out of the data that's coming out of very quickly. Precisely, that's exactly right. Yeah. Which just balances the system out and allows that scale very low latency. And it does it the same way Memcache D has been doing it successfully across some of the largest sites in the world for the last 10 years, right? It uses an LRU mechanism where as one layer in the tier fills up, the least recently used thing gets ejected so the newest, freshest data can come in but it all lives on and each layer does the same sort of behavior. Up, SSD full, let's find the oldest thing that we needed and let it live on just in this. It's really been a lot of innovation.