SCIENCE
IBM Validates Obsidian's Longbow for DB2 pureScale Stretch Clusters
“Mission Critical” Solution for Disaster Recovery and Business Continuity
Following several months of system testing, Obsidian is pleased to announce that IBM has now validated Longbow configurations for its flagship database product, DB2 pureScale. Matt Huras, IBM’s Chief Architect for DB2 pureScale, included material on the “Geographically Dispersed DB2 PureScale Cluster” (or GDPC for short) in his presentation to a receptive audience of knowledgeable network administrators at the International DB2 User Group held earlier this month. Complete details on the solution can be found in an IBM white paper (URL listed below).
DB2 pureScale is a clustered database implementation capable of scaling from small footprints up to very large installations which combine the power of many servers to deliver world-class database performance and mission critical reliability. DB2 pureScale's servers are interconnected with InfiniBand technology, which has historically prevented DB2 pureScale from offering multi-site disaster tolerance. Obsidian and IBM have now removed this restriction with the announcement of GDPC.http://www.ibm.com/developerworks/data/library/long/dm-1104purescalegdpc/index.html
“Originally developed for military networks, Longbows are quickly hitting their stride in the Enterprise Data Center”, said Dr. David Southwell, Obsidian's CVO, “...for example by enhancing DB2 pureScale's value proposition with synchronous multi-site disaster tolerance over metro-area networks. DB2 pureScale's Longbow-enabled GDPC now delivers Enterprise grade scalability, security and availability”.
GDPC distributes pureScale's servers and underlying mirrored GPFS storage across remote sites, connected by Obsidian's Longbows which provide AES-protected InfiniBand links with transparency, low-latency and high-bandwidth. For the first time, this ultra-fast connection allows each half of the cluster to maintain perfect database synchronization across metro-area distances without significant performance penalty.
In the event of a site-wide failure, the surviving site always has a perfect image of the database, which not only continuously protects against data loss but also enables the Enterprise to fail-over database operations to the remaining site thereby maintaining business continuity in the face of major outages. Multiple Longbow links may be established to provide protection against link failure and yield incremental performance benefits.