Archives of the TeradataForum

Message Posted: Thu, 24 Oct 2002 @ 18:37:18 GMT

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

I'm not sure that having execution times (wall-clock times) for any given query has any relevance on the Teradata. You could record stand-alone times, but how does that relate to execution times under various workloads? Any single query can have a major impact on the whole system.

Even in a stand-alone environment, you will still find that execution times vary. If nothing else, remember that there are background tasks performed by the DBMS (for example, dictionary cache spoiling) that can have a significant impact on quick running queries. The quicker your query, the more of an influence these background tasks will have on the execution time of your query.

Also remember that data demographics will have a significant influence on how your queries will execute. For example, a query for one item may return an answer-set faster than the same query that has been submitted for another item.

If you're attempting to compare the amount of work that one query represents against another query, you would be better-off using the contents of column CPUTIME from View DBC.AMPUSAGE (or the underlying system table DBC.ACCTG). If you're interested in that approach, then you might want to read-up on Account String Expansion (ASE).

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