News

Red Hat Updates JBoss Data Grid

• By John K. Waters
• April 10, 2013

Red Hat this week released a new version of its in-memory database JBoss Data Grid.

Built on the open-source, Java-based Infinispan grid computing platform, JBoss Data Grid 6.1 was, according to the company, developed for the cloud and provides a scalable cache that reduces response times and increases reliability for the applications using it. Red Hat describes this release of the data grid as an "intelligent distributed caching solution." It supports Java, C#, and both the Spring and .NET frameworks.

The list of new features in this release includes cross-data center replication (XDR), which allows for replication across clusters and locations. The result, the company says, is better performance for remote users and applications. Another new feature, non-blocking state transfer (NBST), improves elasticity through a new pause-less join-and-leave protocol. This release also comes with improvements to the map/reduce framework, which provides "increased support for long-running compute applications and improved ease-of-use," the company says. It also allows use of a compute grid with a de facto standard programming API.

This release also adds rolling upgrades for an optimized Infinispan-specific protocol called Hot Rod, and support for Context Dependency Injection (CDI), a Java standard for dependency-injection-based module configuration at runtime, which Red Had pioneered in Java EE 6.

An in-memory database stores partitioned data in main memory, rather than on a disk, which puts it much closer to the applications using those data. That proximity cranks the speed of the database (no seek time, fewer CPU instructions, simpler algorithms). Build it on a grid computing system like Infinispan, and you add the ability to manage large amounts of distributed data.

Infinispan, which is a JBoss Community project, ships with a language-independent server module that supports the memcached protocol and almost every popular programming language. Because Infinispan has a REST API, it works not only with Java or JVM-based apps, but virtually any website or application that.

Infinispan supports Temporary Caching for Java, which is currently being developed by the Java Community Process (JCP) as Java Specification Request (JSR) 107 (better known as JCache). In February it was announced that the long-awaited specs for the standardization of temporary, in-memory caching of Java objects wouldn't be included with the upcoming release of Java EE 7. Although Infinispan doesn't currently implement JSR 107 in its current incarnation, work is underway to achieve JSR 107 compliance once the JSR completes.

Red Hat launched the first version of its Data Grid product in June 2012 (it debuted as version 6). The company billed it as a way for enterprises to "reduce the need for relational databases and scale application development for better big data management."

In a blog post announcing this week's 6.1 release, the Red Hat JBoss Middleware Team claimed their part in pushing in-memory data grid technologies into the spotlight. "We've been a part of this rising tide since unveiling Red Hat JBoss Data Grid last year," they wrote.

"We like to think of Red Hat JBoss Data Grid as a supercharger for enterprise applications," the team wrote. "In essence, it takes traditional approaches to application data and turns them on their heads by boosting the speed and reach of data access for applications throughout the enterprise on an as-needed basis - attributes that are becoming increasingly important for organizations that rely on structured or unstructured data for operational and competitive advantages."

More information about the new Data Grid is available on the Red Hat Web site here.