On Wed, 2006-03-29 at 12:19, Antimon wrote:
> Hi,
> I'm sorry about being understood like i was trying to compare pgsql
> with mysql. I was trying stuff, did this and saw that huge difference
> (even it is not bad alone, but comparing to mysql), and thought that
> might be some library issue causing slow reads from server. I don't
> need any rdbms to be faster on selecting a constant integer. My point
> was the library, not server performance.
No need to apologize, honest. This discussion brings up some valid
points, even if the benchmark is not necessarily a valid method for
choosing the database.
PostgreSQL is generally "heavier" than MySQL. A large portion of this
is that PostgreSQL is generally busy making sure things are done right
first, and fast second. In MySQL, it's the other way around.
I wonder how your test would work if you did something a little more
complex (like the pick 1 row in a million scenario) and did in parallel
for say 5, 10, 20, 50 clients at a time. That would like give you some
idea of how well connectivity and small sql select statements scale on
each system.
Generally, pgsql folks consider the single user scenario to not be that
important, and ask themselves "so what happens when 50 people do this at
the same time?" Again, MySQL tends to be the opposite.