Archives of the TeradataForum
Message Posted: Mon, 26 Aug 2013 @ 12:26:51 GMT
<-- Anonymously Posted: Monday, August 26, 2013 05:54 -->
We are running a 4 node 2580 appliance and running short of system resources. By using compression, I have been able to demonstrate a reduction of space for some databases upto 50%. I was comparing simple select queries on compressed tables vs uncompressed ones on dbql. Though the perm space and I/O consumed is far less in a compressed table, the CPU consumed also increases with the degree compression.
(1) I checked some of the manuals and white papers and they seem to suggest that 'the increase in CPU is more than compensated by the decrease in I/O'.
Can you please demystify this for me please? Does it mean that the table (or a part of it) will be available much earlier for spooling than the uncompressed one and therefore available earlier for the CPU to uncompress?
(2) Are there any cases where you folks have decided against compression based on the CPU usage for select? The manuals seem to imply that there could be a performance degrade in the case of INS-UPD-DEL but there will be improvement when it comes to SELECT.
Can you kindly share your thoughts?
|Copyright 2016 - All Rights Reserved|
|Last Modified: 27 Dec 2016|