Couchbase Takes On Oracle, MongoDB And Cassandra

"We're beating up on Oracle, MySQL and PostgreSQL, and, yes, we're having a lot of success in greenfield opportunities against MongoDB and Cassandra," says Couchbase CEO Bob Wiederhold.

What is it about fast-growing startups that breeds these sorts of cocky statements (see Cloudera and Cassandra)?

Wiederhold's comment came on the heels of Wednesday's Couchbase 2.1 release, which brings incremental embellishments to much more extensive upgrades delivered with December's 2.0 release.

Couchbase 2.0 has been "extremely successful," according to Wiederhold, driving "blowout" first- and second-quarter results for the company in part because it added crucial capabilities provided by NoSQL competitors.

[ Want more bold claims from big data upstarts? Read Big Data Pioneers Get Big For Their Britches. ]

"We extended the capabilities of our key value database to also be a document database that could do indexing and querying, and we also introduced cross-data-center replication," he explained. Document-handling and cross-data-center replication (for global deployment) are the strong suits of MongoDB and Cassandra, respectively.

As for the competition with Oracle and other relational databases, those products just weren't designed for the modern era, Wiederhold insists.

"If you were developing an application from around 1995 to 2002, you had no choice but to develop it with Oracle, DB2 or Microsoft SQL Server," Wiederhold told InformationWeek in an interview this week. "Now they're having a difficult time scaling and they're not getting the performance they need with relational databases. They want a horizontally scalable application tier and a horizontally scalable database tier because their mobile and Web applications are very data-centric, and they're connected to the Internet and potentially billions of users."

With the 2.1 release announced this week, Couchbase has added refinements including a multi-threaded persistence engine. That's a mouthful, but it solves the problem of keeping up with random data requests whereby it's hard to predict what data to cache. In these cases, performance depends on how quickly you can get data off disk. The multi-threaded persistence engine basically provides data readers and writers that use all the capacity of multi-threaded compute cores.

Couchbase 2.1 bolsters multi-data center replication with a battery of optimizations. These help global operations ensure performance by serving customers from the closest possible data center. The optimizations minimize mirroring latencies, thereby improving response times.

There's also a new health-check tool that monitors resources including memory, CPU power and disk capacity. As you start scaling up, resources can get out of whack, depending on query and memory demands. The health-check tool spots these problems and suggests how to best alleviate the bottlenecks.

Finally, Couchbase has improved its administrative interface with progress indicators for long-running tasks such as rebalancing operations. It's a small advance, but it puts administrators at ease when they can see progress and know that there aren't problems cropping up in a process.

As we're seeing with many relatively young products, there's a lot of innovation and excitement about new possibilities, but it takes time to apply the finishing touches that add up to maturity.