[rt-users] Speeding up RT3

[Vivek Khera]

On Oct 3, 2004, at 11:02 PM, Jesse Vincent wrote:

> On a large Pg site that I've got ties to, we didn't see a whole lot of 
> change with em, actually.
>

Shocked I am... most of the indexes as defined in the distribution are 
just not used at all.  They can't be since the queries are against 
lower(column) and the index is not.  If you "explain" or "explain 
analyze" the generated queries you'll see what I mean.

However, some of the queries sped up from say 200ms to 85ms, so it may 
not be that noticeable to the average user... :-)  Pg is just plain 
wicked fast anyhow.

I still do see a massive improvement on the "Home" page load time (I 
have 8 queues visible to my personal account).  Unfortunately the 
Debug=1 parameter doesn't cascade thru to the links within the page so 
timing is hard to get accurate -- when you reload the page with 
?Debug=1 appended chances of the data being in the disk buffer cache is 
high so the speed will be way faster.

Try this experiment: turn on the postgres query logger (in 
postgres.conf, set log_min_duration=0), then hit some pages.  Look at 
the generated queries and run "explain analyze" against them after some 
time such that the disk buffers may have flushed out.

The old indexes are never used unless you happen to hit upon an integer 
index, or some of the handful of queries against some ACLs that are not 
lower()'d.


Vivek Khera, Ph.D.
+1-301-869-4449 x806

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 2476 bytes
Desc: not available
URL: <http://lists.bestpractical.com/pipermail/rt-users/attachments/20041006/787c0769/attachment.bin>