Failover database reuse not for all
In this week's issue of GCN, we did an article
about how failover database could be put to work doing additional tasks, such as generating reports.
For many organizations, reusing failover database servers makes darn good sense. Having a failover database system set aside only for disaster recovery is "like keeping a second car in your garage that you never use except when your primary car fails," Juan Loaiza, senior vice president of database development for Oracle, told us. "It's just sitting there collecting dust."
It should be worth nothing, however not everybody we spoke with thought this was a good idea.
Jack Bradham, a technology specialist for Microsoft Federal feels that backup databases should be used for just that ... backups. "Backups should be used for what they are intended for ' a backup of the database for disaster recovery," he said.
He suspects that the major reason that most organizations reuse failover databases is as a way to save costs. Yet this approach may be misguided. Once an organization has invested in a database system and backup system, setting up a third instance is not that much more expensive, between the hardware, additional database software license and replication software. "I don't see the advantage in replicating through backup, if you have a high-quality replication solution," he said.
So good point ' reusing failover databases sounds like a clever idea, but check to see if it really saves you that much money in the long run.
It's also worth noting that even if you don't have a live failover database to query from, you can still make use of a backup database system, or even an archiving system. We got this tip from Craig Mullins, corporate technologist for Neon Enterprise Software. Neon offers backup software for backing up databases to a variety of back-end storage systems, even tape systems. One of the interesting capabilities of its software is the ability to answer calls made from the Java Database Connectivity connectors, in effect allowing Java-based business intelligence systems to query the backup or archive as if it were a full database system.
Of course, the backup system will not respond as quickly as a live database. And depending on how often the database is backed up, it may not have the most-up-to-the-minute data. But neither of these factors may even be considered drawbacks in most cases. And the approach can provide data for reports without relying on the chief database, Mullins said.
Posted by Joab Jackson on Jun 25, 2007 at 9:39 AM