Dave Vellante and John Furrier interviewed Clustrix CTO Aaron Passey on theCube last week. Clustrix offers a highly scalable relational database that’s fully compatible with MySQL. Passey claims Clustrix can scale near-linearly.

The most interesting bit of the interview is when asked about in-memory databases, Passey says that in-memory is basically a hack to that tries to solve the problem of scalability by throwing hardware at the problem. Although Clustrix takes advantage of memory caching, it’s not the only thing it relies on. It’s a stark reminder that there are other ways of dealing with issues of scalability.

Passey emphasizes that typically if you go with an in-memory system you’re trading scalability for persistence. He says you don’t have to make that trade-off with Clustrix. Passey talked about how Clusterix co-founder Sergei Tsarev did some testing comparing MongoDB and Clustrix and found that Clustrix scaled better than MongoDB, but admits that the MongoDB implementation might not have been optimal. But he emphasized that you don’t have to give up any of the features of MySQL to scale with Clustrix. You can even drop Clustrix in as a replacement for MySQL.

Instead scrapping the relational database model in favor of an in-memory, distributed and non-relationol architecture Passey and company decided to fix MySQL. The company is part of the NewSQL movement, which seeks to build highly scalable relational databases. Other NewSQL companies include: Drizzle, ReThinkDB and VoltDB.

Passey says the team decided to start Clustrix while working for Isilon, a storage company that EMC acquired in 2010 for $2 billion. He says that he kept hearing from customers that they wished Isilon had a database solution.

Clustrix’s customers include Photobox, EBI, The Ladders, iOffer.

Services Angle

NoSQL has been a hot topic for the past few years, but huge numbers of systems of record still depend on relational databases and that’s not going to change. Some Web applications and big data problems require the power that distributed non-relational databases, often holding data in-memory, require. But many, probably most, are still best served by the relational model. This is possibly best demonstrated by some of the biggest database announcements of the week. First, Oracle announced that it is releasing its own NoSQL product. But it’s also offering a hosted version of its flagship relational database. And last week, Google announced a hosted MySQL service for its App Engine customers. Google has offered its own NoSQL data store for App Engine customers, but there has been a big demand for relational databases – hence the new Google Coud SQL service. Relational databases aren’t going anywhere, and can still be improved. That’s where companies like Clustrix come in.