RSS Feed

Method R Discussion Group

I have people telling me that if we improve the stats gathering process, performance will improve. What does that mean?

By increasing the sample size of the stats gathering process and or by increasing the frequency of the stats gathering process, it is likely the optimizer will get better data about the underlying objects to calculate cardinality more accurately and hence drive better execution plans. But you need to have evidence that cardinality estimates are way off because your data is changing faster than the stats gathering process or the sample size is way off.