Archives of the TeradataForum

Message Posted: Tue, 08 Nov 2005 @ 21:38:16 GMT

	<Prev	Next>		<<First	<Prev	Next>	Last>>

" The reason that "cluster backups" were mentioned previously is that one can run multiple backup jobs in parallel, each backup job archiving a section or "cluster" of the Teradata configuration, writing to multiple backup devices in parallel. Thus the aggregate throughput of the target media is increased to match the capabilities of the Teradata configuration." - John Grass

That is absolutely correct, which allows for much greater performance. As an additional note to my previous post. In order to achieve parallel restore of a cluster archive, one must use a unique (different) userid for each archive created in the original backup. So if you used five streams to backup 50 clusters (each stream backed up 10 clusters), then you would need 5 userids { bar1, bar2, bar3, bar4, bar5 }, one for each restore stream. You must make sure that the final BUILD command is not performed until all the cluster data has been restored!

Michael E. McBride, MSCIS

	<Prev	Next>		<<First	<Prev	Next>	Last>>