VoltDB expands into Web operations
- 22 January, 2013 15:10
Responding to a growing user base of online service providers, VoltDB has outfitted its namesake in-memory database management system with additional tools to communicate with other technologies usually found in a Web applications stack.
"We're seeing adoption in the Web space, not only for online retail, but for customers like AOL and Yahoo," said VoltDB President and CEO Bruce Reading. "They are using VoltDB to provide a more effective Web performance, to speed up their applications."
Introduced in 2010, the VoltDB database was designed specifically for handling high-throughput transactions that also required all the ACID properties offered by standard relational databases -- ACID stands for atomicity, consistency, isolation and durability.
VoltDB is an in-memory database, meaning the entire database resides in the working memory of one or multiple servers. It was designed in part by Michael Stonebraker, who also helped create the Ingres and PostgreSQL databases. Stonebraker, a cofounder of VoltDB, now serves as chief technology officer for the company. The design of the database offers performance advantages over other in-memory databases, such as Oracle's TimesTen, Stonebraker argued, in that it reduces the need for record-level locking and multithread management.
VoltDB, the database, is now used by almost 200 organizations, according to the company. Originally, finance firms took an interest in the technology, though the user base has been expanding to Web service providers as well, Reading said. The quick reaction times of the database speeds an Internet service's response times for its users.
In addition to the Web connectors, the new version of VoltDB also comes with a number of tweaks that should further boost performance and keep the database online for longer periods of time.
A database schema can now be updated without the need to restart the software, a potentially valuable feature for companies that need to keep their databases running around the clock. A failed node, once fixed, can now rejoin the cluster without downtime. The package also includes a new, speedier, export function, one that, the company claims, can move data to analytic systems 20 times faster than before.
In the new version, VoltDB's palette of SQL commands has been expanded. It now supports for the SQL LIKE, NOT LIKE and UNION calls. It has a number of new column functions too, including those for counting the number of characters within a column field and for indexing queries. "If you have a very popular query, you would like to be able to index it, so it would go fast," Stonebraker said. Ad-Hoc SQL queries can be executed faster -- the database can now run up to tens of thousands of ad-hoc queries per second.
Released as open source under a Gnu GPL version 3 license, VoltDB is available at no cost. VoltDB also offers a commercially supported introductory version for slightly less than US$10,000, as well as an enterprise version that starts at around $35,000.