On 10-07-29 08:54 PM, Greg Smith wrote:
> Brad Nicholson wrote:
>> Postgres also had a reputation of being slow compared to MySQL.
>> This was due to a lot of really poor MySQL vs Postgres benchmarks
>> floating around in the early 2000's.
>
> I think more of those were fair than you're giving them credit for.
I'm sure some where, but I recall a lot that were not.
The main problems I recall is that they took the stock postgresql.conf
(which was far to restrictive) and measured it against a much better
MySQL config. They then measured some unrealistic test for most
applications and declared MySQL the clear winner for everything and
Postgres slow as a dog.
It's one thing for database folks to look at that see the problems
and/or limitations with those sorts of tests. But a lot of developers
were taking these to heart and siding with MySQL and slagging Postgres
as being slow - often unjustly.
> For many common loads, up until PG 8.1 came out--November 8.1--MySQL
> really was faster. That was the release with the killer read
> scalability improvements, then 8.3 piled on again with all the
> write-heavy stuff too. MySQL 4 vs. PG 8.0? MySQL won that fair and
> square sometimes.
>
Oh, I agree that MySQL was faster for some stuff, but not everything.
Back in those days, I routinely saw web sites backed by MySQL 3.x
(forget the exact version) grind to an absolute halt under concurrent
access due to table level locking in MyISAM. Moving those over to the
earlier branches of 7.x improved things drastically.
That said, I also saw the opposite, where MySQL was a lot faster than
Postgres.
--
Brad Nicholson 416-673-4106
Database Administrator, Afilias Canada Corp.