> I've wondered and am still wondering what a lot of these benchmark tests
> are out to prove.
In this case, the "benchmark test" was not out to prove anything. It was
an good-faith result of a porting effort with a suprising (to the
tester) result.
> I'm not sure that any PostgreSQL advocate has ever said or
> implied that PostgreSQL is faster than anything, much less MySQL. While I'm
> sure it's faster than some, I've just never heard the argument for using
> PostgreSQL as "It's faster than anything else".
Very true. But it turns out that in at least some real-world tests, in
this case a real application *built for MySQL*, PostgreSQL was
substantially faster when the number of users climbed above a very small
number. These results are consistant with and supported by GB's initial
published benchmark results.
Two separate styles of comparisons with consistant results might help
someone choose the right solution for their application. No harm in
that, eh?
> I chose PostgreSQL by what
> it could do, not how fast it can SELECT... No benchmark between MySQL and
> PostgreSQL (or any other RDBMS ) is ever going to be truly accurate since
> there are so many things MySQL simply can't to that PostgreSQL (and others)
> can..
Well, that is another dimension to the evaluation/comparison. But the
testing results stand on their own: you *can* choose PostgreSQL for its
performance, and you *will* have made the right choice. This is
especially gratifying for all of us who have contributed to PostgreSQL
because we *didn't* benchmark it, and *assumed* that MySQL claims for
superior speed under all circumstances were accurate. Turns out it may
be true for single-user mode, but that we've built a darn fast RDBMS for
real-world applications.
One *very unfair* part of these benchmarks and comparisons is that both
MySQL and PostgreSQL can be identified by name for the comparisons, so
they tend to be talked about the most. But the GB benchmarks could lead
one to conclude that if SourceForge had been built with another database
product it would also have seen a performance improvement when tested
with PostgreSQL.
- Thomas
