Keeping a database running at all costs can be painful. Traditional approaches to database high availability (HA), such as running Veritas Cluster Server or Microsoft Cluster Server, require IT managers to double their administrative duties, and then some. "Everything you do to the primary server, you have to do to the secondary server," says Steve Norall, director of marketing at PolyServe, Beaverton, OR, maker of the PolyServe Matrix...
Server clustered file system for Linux and Windows.
Traditional HA approaches can also take a long time to get a secondary server online. A failure first needs to be detected and then the secondary server has to claim the disk through a SCSI reserve/release command; then it must check and restore the file system, mount the file system and, finally, load Oracle. This can easily take 20 to 30 minutes. "It's by no means instantaneous," Norall says.
PolyServe's solution to the database HA problem relies, not surprisingly, on its clustered file system software. It works like this: Run Matrix Server on both your primary and secondary database servers, and load Oracle on the same shared storage area network (SAN) volume. Upon detecting that the primary server has failed, the secondary server simply needs to load Oracle; because it sees the same file system as the primary server, it can bypass the restoring and mounting steps.
Norall estimates that PolyServe's database HA package can get you back up and running in under five minutes.
All this came together last month, when PolyServe announced support for Oracle 9i and 10G (it previously only supported Oracle Real Application Cluster), as well as heartbeat monitoring and failover software. The software supports Windows and Linux, and is priced at $3,000 per processor.
PolyServe's approach is a boon to enterprises "moving toward lower-cost systems," such as running Linux on Intel x86 platforms, but that "still want to achieve the same levels of resilience," says William Hurley, senior analyst, Enterprise Strategy Group.