Product of the Month: VoltDB 4.0, from VoltDB Inc.
Release date: Jan. 29, 2014
What it does
VoltDB is an in-memory database focused on large-scale distributed applications. It's part of an emerging category of NewSQL technologies that combine the caching and processing scalability of NoSQL databases with the data reliability and transaction integrity of mainstream relational software based on SQL. VoltDB automatically manages database sharding across shared-nothing server clusters to speed performance while also providing full ACID compliance for guaranteeing transactions. Version 4.0 expands VoltDB beyond its initial emphasis on operational processing by enhancing query throughput capabilities for analytical jobs.
Why it matters
Running transactions and analytical queries on data stored in a system's memory can accelerate processing compared to using disk-based data. Vendor VoltDB Inc. said that as part of its drive to boost analytics performance, VoltDB 4.0 can read data on queries 10 times faster than Version 3.0 could. The database now also supports a larger number of concurrent users in analytics applications, and it includes a variety of new SQL functions for writing queries. In addition, an elastic scaling feature in VoltDB 4.0 lets users add nodes to clusters while they're running. And the new version broadens the language support offered to database programmers beyond Java -- stored procedures can now be written in the Groovy programming language as well.
What users say
VoltDB supports fast data caching, said Marc Firenze, chief technology officer at Eagle Investment Systems LLC, a subsidiary of financial services firm BNY Mellon that uses the database to power cloud-based software it offers to other companies for tracking the performance of investment portfolios and analyzing performance risks. Caching speed is important to Eagle because "calculations need to run at a larger scale in a cloud-based environment" with large and quickly changing data sets, he said.
Firenze said the elasticity improvements in VoltDB 4.0 allow a working cluster to be expanded to meet new customer requirements and upgraded service-level agreements without disrupting ongoing computations and analytics. And the additions to VoltDB's SQL repertoire move the database toward greater compatibility with the SQL-92 version of the language; such compatibility assists in migrating existing SQL functions to a VoltDB cluster, he said.
The cluster elasticity feature is also useful to VoltDB user Openet, a network transaction management software vendor in Dublin, Ireland. That capability enables the company to better support the needs of telecom operators to expand networks and handle more data, according to Alan Carbery, executive director of engineering architecture at Openet.
VoltDB 4.0 "allows us to scale up and down in a cloud environment easily," he said. "This is key for operators who need to constantly expand their networks to handle the growing demand for data."
Openet turned to NoSQL approaches several years ago as part of an attempt to scale its systems to run on large clusters, Carbery said. But he added that disk-based databases required a high-performance storage area network (SAN) to achieve Openet's performance targets, largely due to transaction logging that must be written to disk synchronously. VoltDB's in-memory architecture let Carbery and his team remove the SAN from the cluster architecture.
- Tenfold increase in analytical querying performance.
- Elastic scaling capability for expanding clusters without any processing stoppages.
- Increased support for SQL language functions and materialized views.
VoltDB software is licensed on an annual subscription basis; the company offers startup pricing that begins at $3,500 for a single node, then scales up depending on the number of deployed cluster nodes as well as a variety of performance-related factors.