Thanks everyone for the input so far, Merlin's comment about the network gave me one of those duh moments since I have been running these queries remotely using pgadmin. I will experiment with this more tomorrow/Monday along with the other suggestions that have been posted to hopefully narrow it down. Running the query from my webserver yielded much better times, but from a quick look it seems my 8.4 server is still a bit slower. I will share more details as I dig into it more tomorrow or Monday.
-Patrick
----- Original Message -----
From: "Merlin Moncure" <mmoncure@xxxxxxxxx>
To: "Patrick Donlin" <pdonlin@xxxxxxxxx>
Cc: "Kevin Grittner" <Kevin.Grittner@xxxxxxxxxxxx>, pgsql-performance@xxxxxxxxxxxxxx
Sent: Thursday, July 15, 2010 12:04:13 PM GMT -05:00 US/Canada Eastern
Subject: Re: Identical query slower on 8.4 vs 8.3
your plans are identical as is the runtime basically. this means you
might want to consider the following possibilities:
*) operator error :-)
*) cache effects
*) environmental factors on the server at the time
*) network/client issues
I say network issues because if your explain analyze (which actually
does run the entire query) is significantly faster than the full
query, then we have to consider that the formatting and transfer of
the data back to the client (even if it's on the same box) becomes
suspicious. If you've eliminated other possibilities, try running
other big, trivially planned, mucho result returning queries (like
select * from table) on both servers and comparing times.
merlin
-Patrick
----- Original Message -----
From: "Merlin Moncure" <mmoncure@xxxxxxxxx>
To: "Patrick Donlin" <pdonlin@xxxxxxxxx>
Cc: "Kevin Grittner" <Kevin.Grittner@xxxxxxxxxxxx>, pgsql-performance@xxxxxxxxxxxxxx
Sent: Thursday, July 15, 2010 12:04:13 PM GMT -05:00 US/Canada Eastern
Subject: Re: Identical query slower on 8.4 vs 8.3
your plans are identical as is the runtime basically. this means you
might want to consider the following possibilities:
*) operator error :-)
*) cache effects
*) environmental factors on the server at the time
*) network/client issues
I say network issues because if your explain analyze (which actually
does run the entire query) is significantly faster than the full
query, then we have to consider that the formatting and transfer of
the data back to the client (even if it's on the same box) becomes
suspicious. If you've eliminated other possibilities, try running
other big, trivially planned, mucho result returning queries (like
select * from table) on both servers and comparing times.
merlin
