[pgpool-general: 3351] Optimization problem

[Michal Šťava]

Hi,

i have a question about optimization of pgpool-II.

I have 4 servers:
1] pgpool-II server
2] master db
3] slave01 db
4] slave02 db

I have master/slave + loadbalancing (same priority for each server)

And also i have:
1] postgresql server (same version 9.3 like in pgpool)

INFO: every node is on the same local network and has the same parameters:
1 core, 4GB RAM, postgres 9.3 and java 1.7, application is in java

I have 2 specific scenarios for testing (maybe db cache can be the problem):
1] 50000 values from 1 table in 1 query
2] 50000 values from 1 table in 50000 queries

For scenario 1 I am using 2 different callings:
1] If I am calling pgpool server 10x by my application (so some of them are
on master, some on slave1 and some on slave2) I get time about 2:15. (time
when all processes end)
2] If I am calling one postgreSQL server 10x by my application (so all
processes are on 1 DB), i get time about 1:45.

For scenarion 2:
1] Calling pgpool server 10x by my application, time is 13:39 +- for all
2] Calling postgresql server 10x by my application, time is 6:50 +- for all

Do you have some idea how is that possible and what i can do to optimize
pgpool-II to be better in distribution of read queries than 1 postgreSQL DB?
Maybe I have something bad in settings of pgpool-II server.

Thanks for any advice,
Michal Stava

PS: pgpool.conf in attachment
in pgpool log there is debug info so maybe this can be the problem, but
debug is set on 0?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.sraoss.jp/pipermail/pgpool-general/attachments/20141205/013b5de2/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: pgpool.conf
Type: application/octet-stream
Size: 31248 bytes
Desc: not available
URL: <http://www.sraoss.jp/pipermail/pgpool-general/attachments/20141205/013b5de2/attachment-0001.obj>