Hi Will,
IMHO'.
>Question 1.
>Is the result set for a particular query just stored in a temp table
>in the connection co-server until all the data is finally retrieved?
Yes, I think so.
>Question 2:.
<snip>
>It seems to me that there needs to be a step 3.5, something like: coservers
>send data back to the connection coserver who finishes the join, sort,
>group by, ... operations. or coservers then pull data from the remote
>co-servers finishing the join operation and then pass the data back to the
>connection server to finish sort, group by operations...
I think so too. It would be evident in the Xplan, if u inspect it.
Normally scans would be in all the co's and the joins/sorts in the primary
co, i guess.
>Question 3:.
>Reason behind this question. I want to understand how this works so I can
>recognize possible causes for slowdowns.
To ensure that the workload is well balanced you could use
onstat -g xmf .In whichX_bytes column indicates the bytes sent.
R_bytes column indicates the bytes recd.
These above values should be ideally similar for all co's.
And if u see large values in X_rtrns, xoffs or xo_cycls , u'd
know that there are problems with xmf.
That just covers the xmf'. u need to also investigate if ur query is
optimized'and any other bottlenecks.
Hope that helps.
Do post your findings ..thanks . -raju
________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com