Re: [GENERAL] PostgreSQL on Windows Paper - Mailing list pgsql-advocacy

From Dave Page
Subject Re: [GENERAL] PostgreSQL on Windows Paper
Date
Msg-id 45E32FB7.1070902@postgresql.org
Whole thread Raw
In response to Re: [GENERAL] PostgreSQL on Windows Paper  (Magnus Hagander <magnus@hagander.net>)
List pgsql-advocacy
Magnus Hagander wrote:
>> - It says that 'one should expect performance on Windows to be lower
>> [because of the per-process architecture], especially where large
>> numbers of small queries are made.' That's not really accurate - it will
>> be slower when there are large numbers of short lived connections. Lots
>> of queries in one connection should be fine though.
>
> I believe ou will still see worse performance, because of at least two
> things: context switching is more expensive (much more), and shared
> memory access appears to be more expensive.
> It will be worse if you have short lived connections, of course.

OK, 'relatively speaking'. I think the important part is the connection
setup time, if only because many web apps may setup new connections for
every page (for example) which is where people often seem to come
unstuck and really see the performance hit.

Regards, Dave.


pgsql-advocacy by date:

Previous
From: Susanne Ebrecht
Date:
Subject: Re: successful booth at Fosdem (Brussels)
Next
From: Florian Weimer
Date:
Subject: Re: Jav Database Performance