By using this site, you agree to our updated Privacy Policy and our Terms of Use. Manage your Cookies Settings.
438,778 Members | 1,074 Online
Bytes IT Community
+ Ask a Question
Need help? Post your question and get tips & solutions from a community of 438,778 IT Pros & Developers. It's quick & easy.

redistribute nodegroup

P: n/a
DB2 v7.2 EEE on AIX

I am trying to redistribute the data for a table that is skewed and
wondering if the output of db2gpmap is optimized or is it just
displaying the current information? I also read some where that LOAD
can be used in analyze mode?

Is there a way to redistribute just one table in the nodegroup? I have
mutiple tables in a nodegroup and i don't see any commands other than
REDISTRIBUTE NODEGROUP

Thanks
Visu

Apr 13 '06 #1
Share this Question
Share on Google+
4 Replies


P: n/a
If u run the db2gpmap it will give you the current P.MAP information. I
am not sure what do u mean by optimized ?
Once you get the P.MAP you can customize it and then run the
RE-DISTRIBUTE command.

cheers.....
Shashi Mannepalli

Apr 13 '06 #2

P: n/a
Thanks for the info Shashi.

Do you know if there is a way to re-distribute just one table and not
all tables in the node group

Apr 13 '06 #3

P: n/a
Ian
Visu wrote:
DB2 v7.2 EEE on AIX

I am trying to redistribute the data for a table that is skewed and
wondering if the output of db2gpmap is optimized or is it just
displaying the current information? I also read some where that LOAD
can be used in analyze mode?

Is there a way to redistribute just one table in the nodegroup? I have
mutiple tables in a nodegroup and i don't see any commands other than
REDISTRIBUTE NODEGROUP


You can run db2atld can be done with MODE=ANALYZE, which will
output a new partition map that will result in even distribution
of the data.

However, you can only redistribute everything in the nodegroup,
because the partition map (pmap) applies to the entire nodegroup.

Theoretically you could create a new nodegroup / tablespace / table
and use a custom pmap -- but this would eliminate the possibility of
colocation on joins (tables being joined have to be in the same
nodegroup to qualify for colocation)

If you have 1 table that is bad, you should validate the partitioning
key that was chosen to make sure it was a good choice.

Apr 14 '06 #4

P: n/a
Thanks Ian. I bielieve the option of validating the partitioning key
and re-loading the data will be the best choice for me.

Apr 14 '06 #5

This discussion thread is closed

Replies have been disabled for this discussion.