Thread: w7 vs linux

w7 vs linux

From
Peter Kroon
Date:
Is pgsql faster on linux?
Currently I've made an installation on W7 and the converted queries are about 3 times slower then on mssql.
There's still some optimization to do tho...but the current results don't look to good.

Re: w7 vs linux

From
raghu ram
Date:


On Fri, Nov 23, 2012 at 3:09 PM, Peter Kroon <plakroon@gmail.com> wrote:
Is pgsql faster on linux?
Currently I've made an installation on W7 and the converted queries are about 3 times slower then on mssql.
There's still some optimization to do tho...but the current results don't look to good.

Below URL provides more information on this topic:

Re: w7 vs linux

From
Craig Ringer
Date:
On 11/23/2012 05:39 PM, Peter Kroon wrote:
Is pgsql faster on linux?
Currently I've made an installation on W7 and the converted queries are about 3 times slower then on mssql.
There's still some optimization to do tho...but the current results don't look to good.
In my experience it's somewhat faster on Linux, but I haven't compared extensively on the same hardware.

It's known to be necessary to set shared_buffers lower on Windows for reasons not yet firmly established.

Since you have provided no information about the configuration or hardware, your question isn't much better than "is A faster than B". What's A? What's B?

-- Craig Ringer                   http://www.2ndQuadrant.com/PostgreSQL Development, 24x7 Support, Training & Services

Re: w7 vs linux

From
Scott Marlowe
Date:
On Fri, Nov 23, 2012 at 2:39 AM, Peter Kroon <plakroon@gmail.com> wrote:
> Is pgsql faster on linux?
> Currently I've made an installation on W7 and the converted queries are
> about 3 times slower then on mssql.
> There's still some optimization to do tho...but the current results don't
> look to good.

Wait first you say pgsql, then you say mssql, so which is it?

also single threaded benchmarks don't really mean a lot in the
relational db world. for instance, let's say that pgsql runs a query
in 100ms with 1 thread, but runs the same query in 110ms with 10
threads.  Meanwhile, a db that runs that query in 50ms in a single
thread, but takes 500ms for 10 threads isn't scaling all that well.

So, what are you trying to do, how are you benchmarking, what kind of
performance is important to you?  The simpler the benchmark, the more
useless it tends to be.