Mark Rae wrote:
> I would say that doing the concurrency tests is probably the most
> important factor in comparing other databases against MySQL, as
> MySQL will almost always win in single-user tests.
>
> E.g. here are some performance figures from tests I have done in the past.
> This is with a 6GB databse on a 4CPU Itanium system running a mixture of
> read-only queries, but it is fairly typical of the behaviour I have seen.
> The Oracle figures also scaled in a similar way to postgres.
>
> Clients 1 2 3 4 6 8 12 16 32 64 128
> -------------------------------------------------------------------------------
> mysql-4.1.1 1.00 1.41 1.34 1.16 0.93 1.03 1.01 1.00 0.94 0.86 0.80
> pg-7.4.1 0.65 1.27 1.90 2.48 2.45 2.50 2.48 2.51 2.49 2.39 2.38
Would be interesting to know about the tuning of the MySQL, I guess that
buffers for indexing and sort is well setup, but what about thread
caching? Knowing that will once in a while you will have a connection
burst you can tell mysql to cache thread so that it can save time next
time it needs them.
--
Robin Ericsson
http://robin.vill.ha.kuddkrig.nu/
