VoltDB - a relational alternative to NoSQL databases

Database pioneer Mike Stonebraker, who last year co-founded VoltDB, LLC, has released a stable version of the VoltDB database system. VoltDB is the result of a joint research project undertaken by Brown University, Yale University, HP Labs and the Massachusetts Institute of Technology (MIT), where Stonebraker, the co-founder of open source databases Postgres and Ingres, holds an adjunct professorship. It's described as a relational database management system which uses Online Transaction Processing (OLTP) and SQL (Structured Query Language).

The database retains the data in RAM or disk partitions and distributes both the data and the SQL engine which processes it, across the server or server cluster's processor cores. VoltDB is designed to ensure transactional consistency while eliminating legacy overhead. In order to conform to the ACID database model, VoltDB uses autonomous thread partitioning, while the data itself is replicated within the cluster. According to VoltDB, LLC, on an Intel Xeon X5550-based Dell PowerEdge R610 cluster the database runs five times faster than NoSQL database Cassandra and 45 times faster than an Oracle relational database. VoltDB also allows almost linear scaling across further clusters.

The company expects VoltDB to be used primarily by the finance sector or companies with a web background which manage large amounts of data and rapidly rising transaction volumes. It is open source and available from the project website under the GNU General Public License (GPL). Stonebraker's company offers a commercial version with prices starting at $15,000 per year for a four-node cluster installation. Future releases of the commercial version will include management and monitoring tools.

See also:

• VoltDB Launches Next-generation Open Source OLTP DBMS, press release from VoltDB.
• Apache Cassandra 0.6 released, a report from The H.
• Cassandra Database accepted as a Apache Top-Level Project, a report from The H.

(crve)