Thread: Trying to figure out pgbench

Trying to figure out pgbench

From
William Yu
Date:
My Dual Core Opteron server came in last week. I tried to do some
benchmarks with pgbench to get some numbers on the difference between
1x1 -> 2x1 -> 2x2 but no matter what I did, I kept getting the same TPS
on all systems. Any hints on what the pgbench parameters I should be using?

In terms of production use, it definitely can handle more load.
Previously, Apache/Perl had to run on a separate server to avoid a ~50%
penalty. Now, the numbers are +15% performance even with Apache/Perl
running on the same box as PostgreSQL. How much more load of course is
what I'd like to quantify.

Re: Trying to figure out pgbench

From
"Mohan, Ross"
Date:
I had a similar experience.

regardless of scaling, etc, I got same results. almost like flags
are not active.

did

pgbench -I template1
and
pgbench -c 10 -t 50 -v -d 1

and played around from there....

This is on IBM pSeries, AIX5.3, PG8.0.2

-----Original Message-----
From: pgsql-performance-owner@postgresql.org [mailto:pgsql-performance-owner@postgresql.org] On Behalf Of William Yu
Sent: Tuesday, June 21, 2005 12:05 PM
To: pgsql-performance@postgresql.org
Subject: [PERFORM] Trying to figure out pgbench


My Dual Core Opteron server came in last week. I tried to do some
benchmarks with pgbench to get some numbers on the difference between
1x1 -> 2x1 -> 2x2 but no matter what I did, I kept getting the same TPS
on all systems. Any hints on what the pgbench parameters I should be using?

In terms of production use, it definitely can handle more load.
Previously, Apache/Perl had to run on a separate server to avoid a ~50%
penalty. Now, the numbers are +15% performance even with Apache/Perl
running on the same box as PostgreSQL. How much more load of course is
what I'd like to quantify.

---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
       choose an index scan if your joining column's datatypes do not
       match

Re: Trying to figure out pgbench

From
William Yu
Date:
I wonder if the -c parameter is truly submitting everything in parallel.
Having 2 telnet sessions up -- 1 doing -c 1 and another doing -c 100 --
I don't see much different in the display speed messages. Perhaps it's
an issue with the telnet console display limiting the command speed. I
thought about piping the output to /dev/null but then the final TPS
results are also piped there. I can try piping output to a file on a
ramdisk maybe.



Mohan, Ross wrote:
> I had a similar experience.
>
> regardless of scaling, etc, I got same results. almost like flags
> are not active.
>
> did
>
> pgbench -I template1
> and
> pgbench -c 10 -t 50 -v -d 1
>
> and played around from there....
>
> This is on IBM pSeries, AIX5.3, PG8.0.2
>
> -----Original Message-----
> From: pgsql-performance-owner@postgresql.org [mailto:pgsql-performance-owner@postgresql.org] On Behalf Of William Yu
> Sent: Tuesday, June 21, 2005 12:05 PM
> To: pgsql-performance@postgresql.org
> Subject: [PERFORM] Trying to figure out pgbench
>
>
> My Dual Core Opteron server came in last week. I tried to do some
> benchmarks with pgbench to get some numbers on the difference between
> 1x1 -> 2x1 -> 2x2 but no matter what I did, I kept getting the same TPS
> on all systems. Any hints on what the pgbench parameters I should be using?
>
> In terms of production use, it definitely can handle more load.
> Previously, Apache/Perl had to run on a separate server to avoid a ~50%
> penalty. Now, the numbers are +15% performance even with Apache/Perl
> running on the same box as PostgreSQL. How much more load of course is
> what I'd like to quantify.
>
> ---------------------------(end of broadcast)---------------------------
> TIP 9: In versions below 8.0, the planner will ignore your desire to
>        choose an index scan if your joining column's datatypes do not
>        match
>
> ---------------------------(end of broadcast)---------------------------
> TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org
>