I wrote a C# (.NET v2) single-threaded console application in Microsoft
Visual Studio 2008, which uses the PostgreSQL ANSI 8.03.04.00 ODBC
driver to connect to PostgreSQL database running on a standalone server.
The application retrieves raw data from the database, performs analysis
and stores results to the database. Multiple instances of the
application can be created at a time and run simultaneously. When
multiple instances of the application are created on a multi-core
machine the performance of the application is much slower than when the
processes are executed on different nodes (e.g., 3 processes on 1
quad-core machine run approx. 25% slower than 3 processes on 3 quad-core
machines). In case of 4 processes the difference is almost 50%. The CPU
load when 3 processes are running on 1 quad-core machine was approx. 75%
(which was as expected).
Each process uses following ODBC connection string:
I did not used any profiler but it seems that the overhead when multiple
processes are running on 1 machine is in incorrect configuration of the
ODBC driver or the driver itself. Any ideas what might be causing the
slowdown? Thanks for your help.
Best regards, Josef
pgsql-odbc by date
|Next:||From: Jonathan Foy||Date: 2010-04-07 19:57:48|
|Subject: Acucobol, ODBC and Postgres|
|Previous:||From: raja agireddy||Date: 2010-03-30 16:56:38|
|Subject: iodbctest: fatal: relocation error|