The NoSQL database market is expected to grow from $200 million in 2013 to $1 billion in 2017, according to Forrester Research, and DataStax has been among the leading support providers along with 10Gen, Couchbase, and Basho (which support MongoDB, Couchbase and Riak, respectively).
Database market share leader Oracle also has a NoSQL database, introduced last year, but it's that company's huge relational database installed base that most NoSQL vendors are going after. DataStax recently claimed that "dozens" of its customers had migrated from Oracle Database or Oracle MySQL to Cassandra in recent quarters. DataStax reports it now has more than 300 customers, including 20 Fortune 100 companies.
[ Want more on ambitious upstart big data vendors? Read Big Data Pioneers Get Big For Their Britches. ]
Cassandra is often seen in so-called "Web scale," global applications with high data volumes and data variety and lots of interactions with customers, mobile devices, sensors or all of the above. Cassandra can scale a single database instance across many machines and multiple data centers with a lower cost to scale than is practical with commercial relational databases.
DataStax Enterprise (DSE) 3.1, the commercial release announced Tuesday, improves on scalability by packing in more data per node. The practical limit per node used to be about 1 terabyte, but DataStax rewrote the storage engine behind DSE 3.1 to increase capacity to 5 to 10 terabytes per node. That makes deployments that much more cost effective, according to the company.
The other big improvement in DSE 3.1 is continued advances in the Cassandra Query Language (CQL), the SQL-like dialect that lets customers work with data the way they're used to with conventional relational databases. Aside from new .NET and Java drivers, CQL now offers better support for repeatable queries.
"CQL gives you select, create index, create table and other statements, and the advance in DSE 3.1 is that CQL lets you partition tables and sort rows within those partitions," said Jonathan Ellis, co-founder and CTO of DataStax, in a phone interview with InformationWeek. "That lets you compute materialized views that Cassandra will keep ordered as you insert and update data so you don't have to do expensive sorts at query time."
As with previous releases, DSE 3.1 delivers more than Cassandra; it also includes Apache Hadoop and Apache Solr, so deployments support batch analytics and enterprise search against the same data managed by the NoSQL database. OpsCenter is the DataStax monitoring and management software for running the entire deployment.
DataStax also announced DataStax Community Edition (DSC) 2.0 based on Apache Cassandra 2.0, which is set for release next month. DataStax doesn't recommend production use until Cassandra 2.0's formal release in August, but new features include lightweight transaction support that ensures that users don't overwrite the same values, triggers for event-driven database operations, improved compaction for storage efficiency and CQL cursors that make it easier to navigate data sets, according to DataStax.