Re: Linux vs. Windows performance.

Rick Denoire <100.17706_at_germanynet.de> wrote in message news:<tqjq20tcb0j7b1159q6be17uj47ccrebgj_at_4ax.com>...
> Michael Rothwell <marothwellRemoveThis_at_yahoo.com> wrote:
>
> >Has anyone taken an Identical Windows box and Linux box (same hardware
> >configuration) and tested Oracle performance? I'm looking for some real
> >world examples if I move from Windows 2k to linux with Oracle 9i. The
> >box currently has 2 1.8 CPUs with 4 disks and 1 G ram. OLTP system with
> >tables using about 30 G of space. About 40 - 100 concurrent users, but
> >this may double soon. Can I expect better performance from the Linux box?
>
> Someone will perhaps confirm or deny these differences: There is no
> such thing like shared memory in Windows (disadvantage). The lack of
> OS buffering in Windows turns out to be a "built-in" advantage when
> running Oracle. In most cases, one would rather mount file systems
> with the directio option to avoid double buffering. Windows does not
> need to disable what it does not have.
>
> Whenever I ran Oracle on Windows and Linux, these were different
> machines, so I can't really compare. But I have only seen Oracle
> getting unrestartable only in Windows, and exactly only once.
>
> Bye
> Rick Denoire

Rick,

lack of shared memory is not a disadvantage in this case, as it is not needed for win32.
It is just different.

There is a single OS process per oracle instance. As there are not multiple OS processes, why would one need to share memory across them?

Please consult the docs on Metalink for this. This is a solved problem, a non-issue. It is well documented, and just requires you reading the docs.

Pd

<ducking and running>
why is there so much fuss about threading out there these days, if multiple threads under a process are so evil? Received on Fri Feb 13 2004 - 21:50:28 CST