Archives of the TeradataForum

Message Posted: Mon, 29 Feb 2016 @ 09:21:48 GMT

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

Looks like a problem with not enough AWT.

First you can check resusagesawt - columns inusemax and flowctlcnt. InUseMax should be kept on save level (<=62) and flowctl shouldn't be >0.

If not - you can decrease some throttles and/or define expedited AWT for tactical queries.

You can also consider switching DBScontrol parameter EnableExpediteExp to 2 - expedited AWT type 9 will be used by PE(for tactical queries) instead of traditional AWT type 1.

Second issue may be IO - on 2750 you should care rather for IO then for CPU. If you have 30% CPU free you can consider BLC and/or all other kinds of compression to keep used tables as small as possible.

Next step is verification of SPLs -are you sure you must generate 100-120 queries per one MQ Series message? It means over 200 000 queries per hour, what may be to many for 2N2750.

E.g. if you have many delete where PI=... you can consider changing them to one delete wehere PI in (....). It will be still few amps query, but number of parsed queries will be much less.

Another problem which you can analyze are possible locks - check in dbqlsteptbl if some MLK steps have elapsed time >0.

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