On Fri, Feb 20, 2009 at 2:35 PM, Robert Haas
<robertmhaas@gmail.com> wrote:
> First of all, you need to do some research on the benchmark kit itself,
> rather than blindly downloading and using one. BenchmarkSQL has significant
> bugs in it which affect the result. I can say that authoritatively as I
> worked on/with it for quite awhile. Don't trust any result that comes from
> BenchmarkSQL. If you fix the bugs, Oracle (out of the box in OLTP config)
> will come out 60%.
60% what?
Faster than PG 8.3-dev with 100 warehouses (when I last tested it).
> Oracle comes out twice as fast as PG on Linux. And, unless you're using a
> significant number of warehouses, MySQL+InnoDB will come out better than PG
> as well.
I can believe that MySQL could come out faster than PG because I've
had previous experience with it being blindingly fast. Of course I've
also had experience with it having amazingly poor data integrity.
That was MySQL+InnoDB. I haven't really had any integrity problems in that configuration.
I would be pretty surprised if Oracle were in general twice as fast as
PG - what are they doing that much better than what we're doing? I
could certainly imagine it being true in cases that rely on specific
features we lack (e.g. join removal)?
DIO + AIO + multiple DBWR processes + large buffer cache + properly sized logs/log buffers makes a big difference. There are also several other concurrency-related tunables which contribute to it as well.
--
Jonah H. Harris, Senior DBA
myYearbook.com
