John R Pierce wrote:
thats odd, as a domain socket should be slightly -faster- than a tcp/ip
socket. now, since you say 'previous host' I could wonder if other
configuration items are impacting this, such as buffer sizes in
postgresql.conf, relative speed of disk controllers, etc. or perhaps
this new database hasn't been analyzed since it was populated, or its
indexes need rebuilding, or something similar...
I think that is the issue. Previous server was old machine and the
settings in postgresql.conf were the default ones.
The new machine is Xeon quad with 8GB ram and I already made some
changes in postgresql.conf, but still no result.
I followed the instructions on
http://www.powerpostgresql.com/PerfList/ and made the following changes:
shared_buffers = 50000
work_mem = 512000
checkpoint_segments = 32
effective_cache_size = 200000
Also I ran REINDEX on the DB and VACUUM VERBOSE ANALYZE on each
table, but still the results are the same as before.
What are the settings you recommend for this server?
The application that is going to use it has a very demanding back
end, it parses very large XML files (20+) and saves the data in the DB.
Thanks, Kosta
--
Sent via pgsql-general mailing list (pgsql-general@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general
