On 30 Apr 2004 at 7:26, Dennis Bjorklund wrote:
> On Fri, 30 Apr 2004, Gary Doades wrote:
> > I should have also pointed out that MSSQL reported that same index scan
> > as taking 65% of the overall query time. It was just "faster". The
> > overall query took 103ms in MSSQL.
> Are your results based on a single client accessing the database and no
> concurrent updates?
> Would adding more clients, and maybe having some client that
> updates/inserts into the tables, still make mssql faster then pg? Maybe
> it's so simple as pg being optimized for more concurrent users then mssql?
> I'm just asking, I don't know much about the inner workings of
> /Dennis Björklund
At the moment it is difficult to set up many clients for testing concurrent
stuff. In the past I have had several SQLServer clients under test,
mainly select queries. MSSQL can certainly execute queries while other
queries are still running in the background.
Our production app is fairly well biased towards selects. Currently it is
about 70% selects, 20% inserts, 6% deletes and 4% updates. Very few
updates are more than one row based on the primary key. Over 90% of
the time spend running SQL is in select queries.
My limited concurrent testing on Postgres gives very good performance
on updates, inserts, deletes, but it is suffering on the selects in certain
areas which why I have been concentrating my efforts on that area.
Having got similar (or the same) access plans in both Postgres and
MSSQL I was getting down to the next level of checking what was going
on when executing the already planned query.
I do have another database system I could try. Sybase SQLAnywhere.
This is not the original Sybase Entrerprise which has the same roots as
MSSQL. In the past my testing suggested that SQLAnywhere
performance was as godd or better than MSSQL. I mey try to set it up
with the same data in these tests for a more detailed comparison.
---------------------------(end of broadcast)---------------------------
TIP 4: Don't 'kill -9' the postmaster