There’s a new version of VoltDB, the in-memory relational database, with changes including JSON support and speed improvements. The team is reporting sub-millisecond latency for transactions in a relational database.

VoltDB is an in-memory relational database designed by a group of well known database designers, headed by Michael Stonebraker of Ingres and PostrgreSQL. VoltDB is based on the H-Store technology, is ACID-compliant and has been designed to combine low latency and high throughput. VoltDB transactions are created as stored procedures written in a mix of SQL and Java.

According to a blog post about the new version by John Piekos

“We spent much of the past year re-engineering the transaction coordination architecture to minimize the communication  between cluster nodes during transaction processing yet still provide full ACID properties. The result is that version 3.0 executes at a significantly lower latency and is able to execute significantly more transactions in the same time window on the same hardware.”

One advantage of the lessening of inter-node communications is that VoltDB is more cloud friendly.

In addition to the new support for JSON, the SQL support in VoltDB has also been improved. The SQL in earlier versions was somewhat limited, but the new release adds support for
SQL LIKE and NOT LIKE, column functions, indexing column functions, and UNION. You can now also use conversational SQL in addition to stored procedures.

A high performance export feature has also been added that moves data to analytic systems twenty times faster, and there are now export connectors for Netezza, Vertica and JDBC.
New client library languages have also been added, with a new PHP client, Google Go client and version 1.0 of the Node.js client.

Redis - Open Source from Microsoft

New Release of Oracle's TimesTen