Does anybody know how the row count in the execution logs is worked out?I have a report that is made up of a series of datasets obtained from a collection of stored procs. Most of the datasets are only likely to produce row counts in the low single figures and at most the low double figures. One of them is much more open ended but in [i]normal[/i] usage shouldn't produce more than 150 rows. These datasets are pulled together into document that comprises various tables and text boxes that are sent to the police as required. The whole point is for concision and clarity so we don't want to provide huge amounts of raw data.I've been monitoring our report server to try and track an intermittent slowdown and last night there was spike in CXPACKET waits so I had a look at my report usage report(!) and the report that looked like it caused the spike allegedly returned over 600000 rows. This didn't look right so I ran all the individual procs using the same parameters as the report was run with and none of them returned more than 100 rows. The only explanation I can come up with is that the row count on the execution log is calculated by something like Dataset1 x Dataset2 x Dataset3 etc. Can anybody shed any light on this please?
↧