Archives of the TeradataForum
Message Posted: Sun, 21 Feb 2010 @ 15:32:18 GMT
<-- Anonymously Posted: Sunday, February 21, 2010 04:31 -->
I know for a single query, we can calculate the parallel efficiency as following:
ImpactCPU / TotalCPU
But if we calculate this metrics for a user or a system, can we really use
sum(ImpactCPU) / sum(TotalCPU) ?
I cannot understand sum(ImpactCPU). Is it possible that several single queries of one user are skew, but since they are skew on different AMP, so this user's Parallel Efficiency is not that bad? Can somebody clarify?
|Copyright 2016 - All Rights Reserved|
|Last Modified: 28 Jun 2020|