I have a perl test script that performs 1000 random updates to a table.
If I run it on one of our Windows 2000 machines at work, it takes
about a minute to finish a pass. If I run it at home on my linux
machine, it finishes in under two seconds. It didn't seem to matter if
I used a default connect or a tcp connect on linux. I also tried
running against the linux database from another linux machine (486-66
DX2) across a 10Mb ethernet connection. It still completes a pass in
under 6 seconds.
I downloaded and installed the DBD::Pg module to replace DBD::ODBC and
it helped reduce the run time to about 40 seconds on Windows. Both DBD
backends were linked against a LIBPQ.DLL that was built in release mode
with Visual C++ 6.
Both systems are running PostgreSQL 7.1.3. The windows machine has the
latest cygwin and cygipc (just upgraded today). The linux machine was
installed from SuSE 7.2 and PostgreSQL was upgraded using an rpm from
their ftp site. None of the configuration options were changed on
either system.
Does PostgreSQL really run that much faster on linux/unix than on
Windows or is there something I can adjust somewhere to improve the
Windows performance? Are there any tuning or performance related
documents anywhere on the net?
--
tom_p
pfau@nbpfaus.net -- http://www.nbpfaus.net/~pfau/
Crosspointe Community Church information
at http://www.crosspointecommunity.com/
