IBM and Partner Voltaire Inc. last week at the LinuxWorld Expo in New York introduced technologies that will enable enterprises to scale their Linux-based applications running on DB2 databases.
IBMs new DB2 for Linux Clustering solution lets businesses swamped by e-business-generated data scale systems from two to 1,000 nodes, adding nodes as required.
The offering is based on a building-block concept that provides a growth path that can potentially encompass both IBM and non-IBM software.
It works with the Armonk, N.Y., companys WebSphere application server and its Tivoli systems management software, as well as with SAP AGs R/3, MySAP and MySAP Business Intelligence applications, according to IBM officials.
The software is aimed at medium-size enterprises that are looking to grow, including financial, retail, manufacturing and public-sector industries, said Jeff Jones, IBMs director of strategy, Data Management Solutions.
"The rate and amount of data being collected is not slowing down," Jones said. "The drivers are only accelerating, as businesses continue to get real good at collecting all sorts of data. Of course, the cost to collect this data is more urgent than ever."
Meanwhile, Voltaire is offering DB2 users a way to run transaction-heavy applications on cheaper, entry-level servers linked by InfiniBand interconnect technology rather than on larger and more expensive high-end systems.
Voltaire, of Bedford, Mass., demonstrated a 40-node cluster of IBM servers running SAP and IBM applications under the SuSE Inc. Linux Enterprise Server operating system on a DB2 Universal Database. Technology linking the machines included Voltaires flagship nVigor switch router.
The company is expected to release an enhanced version of the technology—which will be renamed InfiniBand Switch Router—later this quarter, officials said. Powering the InfiniBand connection will be 10G-bps chips from Mellanox Corp.
InfiniBand is a channel-based, switched-fabric architecture. Originally predicted to replace other interconnect technologies, it appears that it will at first gain traction in such areas as supercomputing and high-performance clustering.