I have 18 years of mainframe DB2, and I just starting working with UDB LUW
V8.1 FixPack 6 on Sun Solaris last week.
I have been dismayed on the lack of detailed info available in the manuals
on tuning and monitoring REORGS on large tables. I am used to the wealth
of usage notes in mainframeland.
I am recommending the implementation of clustering indexes and wholesale
reorging of several large tables (up to 37GB)/ 200 mill plus rows.
I am familiar with the doc on the "Classic Reorg" and "In Place Reorg" and
I am planning on requesting an outage and using the "Classic Reorg" for
this process, while is going to include the dropping and recreating of
indexes to implement clustering prior to the reorg.
I could really use some work arounds or deeper detail on the following
issues:
1. There is no CREATE INDEX DEFER YES, so I am assuming that the
DROP/RECREATE of index as a clustering index will actually go through the
laborious process of building the index prior to the reorg, and hold a
catalog lock the whole time. True ?? I hope I'm missing something here.
2. There is no -DISPLAY UTILITY command. Familiar with the DB2 LIST
HISTORY command, but this doesnt seem to display anything till the job is
done. Somebody please tell me there's a way to monitor this thing while
its unloading,sorting,reloading,building indexes, waiting on locks, etc..
3. All tablespaces currently exist in one DB on the instance. Does this
mean that all my reorg jobs will be fighting for temp tablespace and sort
resources if I attempt to run concurrently ?
4. WHere are the calculations to figure out the projected temp tabelespace
and sort memory requirements based on the table size and number of
indexes ?
I dont mean to slam UDB, but where the heck is all that information for
doing this sort of thing ?
Thanks, Jeff Kluth
Email: be*********@comcast.net