Re: Is there a meaningful benchmark?

Okay, you've given me some useful information.

As the original subject line indicates, I'm open to the idea that no such benchmark exists.

If anyone asks about this stuff, I can just say that performance varies widely by database and application, that Postgres performs well enough against other RDBMSs, that Postgres is known to scale up well and make good use of concurrency, and that I couldn't find any clear benchmark results to back it up.

Of course, if I *did* find any benchmark values then I could have used that to dispel false rumours from the MySQL guys. However it looks like simple measured indicators aren't easy to come by.

-Will

-----Original Message-----
From: Scott Marlowe [mailto:scott(dot)marlowe(at)gmail(dot)com]
Sent: 19 March 2009 18:14
To: Will Rutherdale (rutherw)
Cc: pgsql-general(at)postgresql(dot)org
Subject: Re: [GENERAL] Is there a meaningful benchmark?

On Thu, Mar 19, 2009 at 3:56 PM, Will Rutherdale (rutherw)
<rutherw(at)cisco(dot)com> wrote:
> I'm having trouble with the tweakers reference below.
>
> I was hoping to see something where hardware platform is held constant while RDBMS is varied, but it seems to be just the opposite.  Or maybe I didn't read the article the right way.

The tweakers test is actually VERY telling. With a constant load, how
well do the database scale as you improve the hardware you've given
them to run on. Hardware is cheap, downtime is not. If you can toss
a 16 core server at a performance problem for $20,000 or so, that's
probably way cheaper than watching your main db chug under load and go
down twice a day.

Conversely, telling the bossman you need that 16 core server to
improve performance and seeing the new server collapse under load
faster than the old one due to poor concurrency is not gonna win you a
lot of brownie points.