Re: Distributions

Jay Aymond wrote: > > What is the difference in running the following two statements: > > (1) update statistics medium for table xzy distributions only; > > or > > (2) update statistics medium for table xyx (col1) distributions only; > update statistics medium for table xyz (col2) distributions only;> .... > for all columns in the table > > ___________________________________________________________ > Jay Aymond > EXE Technologies > jay_aymond@exe.com Hi Jay, 1st) The difference: In (2) you used a different table "xyx" instead of "xyz". The (1) "update statistics ..." might be performed parallel. ( a single table scan, but parallel sort operations ). The (2) "update statistics ..." statements will be performed in sequential mode. The degree of parallelism (1) can be set up by the environment variable "DBUPSPACE" ( the total temporary sort space in KB ). 2nd) Do you think it's a good idea to give the optimizer detailled informations for columns, if the information will not cause a better query plan ? Don't you believe that it will slow down those queries that do not benefit from the detailled distributions ? Be carefull if you start "update statistics medium/high for table xyz". Only make use of "update statistics high/medium ..." when you are sure that the optimizer will need this additional information. Bye Stefan Weideneder