Re: Performance Issues: "order by"

Is the clustered index created before the data was loaded? If so, the rows in the table isn't arranged by the order of the index. You could rearrange it by : alter index INDEX_NAME to cluster; Hope this helps, Zandy In article <6l2gud$a0c$1@geraldo.cc.utexas.edu>, tbryan@arlut.utexas.edu wrote: > > > I've got Informix Online Workgroup Server 7.2 running under Solaris. > The DB will be serving some applications which require that the data > be provided in order based on one of the columns. > > I have a clustered index on that column. When I ran a test table > with an ESQL/C program, I was surprised that it took about > 12 seconds (when the machine was unloaded) during the open cursor > stage. The test table had 70,000 rows of about 55 bytes each. > > Using SET EXPLAIN, I see that the DB is doing a table scan since the > whole table must be returned, but it appears to be spending a lot of > effort ordering the query results into a temporary space. > > If I have a clustered index on this column, would the DB spend time > ordering the data anyway?! Why? Are there any other tricks I can > use to improve the DB performance when dealing with applications > which demand ordered results? > > P.S. Is there any good way to figure out how much time is being > spent just to read the data from disk versus the time spent doing > processing associated with the query? SET EXPLAIN output is rather > terse. > > Thanks. > ---Tom > > -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/ Now offering spam-free web-based newsreading