Method R Discussion GroupMethod R helps developers, DBAs, and decision-makers optimize Oracle-based software in every phase of the software life cycle. |
||
Discuss Method R products, techniques, and events. Ask new questions or post your own responses. |
Do you have a recommendation as to whether to use auto size or skew only size?
The SIZE AUTO setting can be problematic in that it will collect histograms on any "recently used" column (as tracked in the SYS-owned table COL_USAGE$). If a column with data skew that should have a histogram hasn't been used recently enough in a query for it to show up in the table, then the collection will not produce a histogram on that column. However, it will collect histograms on columns like date fields and primary keys that are not skewed.
So, I prefer to use SIZE SKEWONLY to detect skewed column values initially and then follow with subsequent collections using SIZE REPEAT to recollect histograms only on the columns that previously had them. |
Powered by FogBugz