Re: Force Statistics Collection DB2/400?

["Lennon_s_j@xxxxxxxxxxx" <lennon_s_j@xxxxxxxxxxx>]

I wonder if you might be able to force some of the statistics to be 
created by running queries after you restore the file.  For example, a 
SELECT primary key field(s), count(*) type of query might cause some 
rebuilding.

Isn't there something in iSeries Navigator that shows you the 
statistics?  As I recall, a DSPFD over a file will show the number of 
distinct key values so I wonder if there is an equivalent for SQL tables.

Sam

On 4/19/2010 9:04 PM, Ryan Hunt wrote:
> The business data is being copied to tape from V5R4 and restored to V6R1.
> It is then run through an ERP vendor conversion program to move from one
> level to another. The box is an i520 currently split into two LPAR's.
> Memory and processor are identical between the two...although I/O subsystems
> are very different.  Yes, I'm sure the access plans may also been
> recalculated for just about everything.  The reason my eye was on the
> statistics was because disk use (as a proporation) was high (as opposed to
> CPU) and the QDBFSTCCOL process seems to be the culprit.
>
> Either way...seems unavoidable and eventually works itself out.  I'll just
> make sure we do stress testing after all CRP's have gone through a couple of
> rounds, instead of making it a part of the CRP's.
>
> Thanks
>
> Ryan