For those that care :) here's information on the DB query performance testing I was
doing:
For BZ 815341 (
https://bugzilla.redhat.com/show_bug.cgi?id=815341 ) I had to tweek a
couple of our big group queries in order to give us proper availability data (we now can
get information on the UNKNOWN state).
Because this involved some of our more complex and expensive queries, I spent a good
amount of time running some DB tests both with and without my fix on both Postgres and
Oracle. I used really large group sizes (probably larger than the majority of RHQ users
would have in inventory) so for the majority of RHQ users I think they probably won't
see much of any noticeable impact.
See my comment (and the attachment) in the BZ here:
https://bugzilla.redhat.com/show_bug.cgi?id=815341#c13
In short: as expected, it made our slow queries slower :) but it wasn't too bad. We
are querying for more data, so it is to be expected the queries would be slower, but it
doesn't slow it down so much as to make it much worse than what they already are. And
the one query - getResourceGroupComposite - is still overall very fast (which was
surprising to me).
So, I think the fix is good. If anyone wants to look at the data and comment in that BZ,
feel free.