Archives of the TeradataForum
Message Posted: Mon, 07 Apr 2003 @ 17:08:54 GMT
Just an update to clarify a point that came out recently on TDATA, regarding the SESSIONS parameter in ARCMAIN and NTA582 - the tech alert relating to data loss should the SESSIONS parm be coded at a value higher than the AMP configuration.
We believed that data loss also occurred when the SESSIONS count is coded at a value less than the AMP config, when running at a DBS version less than V2R4.01.03.29. However, I've been speaking with Ysbrand Bergman from NCR at length on this point, and it turns out that the tech alert was comprehensive after all. I think the subtle difference in understanding with all of this, is that the AMP config should be regarded as AMP per cluster config, and not the total number of AMPs in the system. So, if you have a 192AMP system across 24 clusters for example - therefore, 8 AMPs per cluster, SESSIONS should be set to a maximum of 8 per single cluster archive script (assuming that your system is evenly balanced). Should the script be covering 2 clusters, then max. SESSIONS = 16.
Pretty obvious in hindsight, but hindsight is a wonderful thing!!
Anyway, just dropping a mail to reassure you all that the original NTA was correct - it was just my understanding that wasn't! Thanks again to Ysbrand for his time, and for getting involved.
|Copyright 2016 - All Rights Reserved|
|Last Modified: 27 Dec 2016|