Hello:
I have installed the same Oracle version (10.x) on a Windows 2003
Server and on a Linux RHAS 3. I have created the same database and run
some querys. The hardware and the data are the same and in Linux I have
set the recommended kernel parameters. The W2K3 filesystem is NTFS, and
the Linux FS is ext3. They are default installations.
The RHAS is three times slower than the W2K3. Why? What can I do to
improve the Linux performance?

Thanks.

<jonas@cyberspace.org> wrote in message
news:1114969234.728397.289910@o13g2000cwo.googlegr oups.com...
> Hello:
> I have installed the same Oracle version (10.x) on a Windows 2003
> Server and on a Linux RHAS 3. I have created the same database and run
> some querys. The hardware and the data are the same and in Linux I have
> set the recommended kernel parameters. The W2K3 filesystem is NTFS, and
> the Linux FS is ext3. They are default installations.
> The RHAS is three times slower than the W2K3. Why? What can I do to
> improve the Linux performance?
>
> Thanks.
>


well why should it be as quick or faster?

jonas@cyberspace.org wrote:

> Hello:
> I have installed the same Oracle version (10.x) on a Windows 2003
> Server and on a Linux RHAS 3. I have created the same database and run
> some querys. The hardware and the data are the same and in Linux I have
> set the recommended kernel parameters. The W2K3 filesystem is NTFS, and
> the Linux FS is ext3. They are default installations.
> The RHAS is three times slower than the W2K3. Why? What can I do to
> improve the Linux performance?
>
> Thanks.

In every benchmark I have ever performed RHAS 3 screams past any version
of Windows. So I am left wondering what you've done wrong.

Did you apply the recommended patches to RHAS? Replace the compiler
before installing the Oracle binaries?

But one possibility is the chip set. Intel, recently, has been burning
Windows specific optimizations into their silicon which makes Windows
faster than Linux. Move to an AMD chip and that isn't the case and is
part of the reason AMD is doing so much better in Oracle shops.
--
Daniel A. Morgan
University of Washington
damorgan@x.washington.edu
(replace 'x' with 'u' to respond)

This was what I have read sometime ...

The system is an Intel Itanium 2, RHAS certified, and the update the 4.
The steps: http://www.puschitz.com/InstallingOracle10g.shtml.
What do you weant to say with 'Replace the compiler
before installing the Oracle binaries?'

Thanks

jonas@cyberspace.org wrote:
> The system is an Intel Itanium 2, RHAS certified, and the update the 4.
> The steps: http://www.puschitz.com/InstallingOracle10g.shtml.
> What do you weant to say with 'Replace the compiler
> before installing the Oracle binaries?'
>
> Thanks

A reference to the following from the installation doc:

Red Hat Enterprise Linux 3:

make-3.79.1
gcc-3.2.3-34
glibc-2.3.2-95.20
compat-db-4.0.14-5
compat-gcc-7.3-2.96.128
compat-gcc-c++-7.3-2.96.128
compat-libstdc++-7.3-2.96.128
compat-libstdc++-devel-7.3-2.96.128
openmotif21-2.1.30-8
setarch-1.3-1
gnome-libs-1.4.1.2.90-34.1

Did you verify you had the correct compiler version.
--
Daniel A. Morgan
University of Washington
damorgan@x.washington.edu
(replace 'x' with 'u' to respond)

jonas@cyberspace.org wrote:

> The RHAS is three times slower than the W2K3. Why? What can I do to
> improve the Linux performance?

Do you use locally attached ATA-disks?
If so, what does hdparm tell you?

Regards,
uwe

<jonas@cyberspace.org> wrote in message
news:1114974589.940984.32590@f14g2000cwb.googlegro ups.com...
> The system is an Intel Itanium 2, RHAS certified, and the update the 4.
> The steps: http://www.puschitz.com/InstallingOracle10g.shtml.
> What do you weant to say with 'Replace the compiler
> before installing the Oracle binaries?'

If you did follow Werner's excellent instructions then, like me, you will
have crippled the performance of the database by not enabling either
asynchronous or direct IO, assuming the filesystem you chose supports them,
and if it doesn't you chose the wrong one. For me this made linux half the
speed of windows on a somewhat naive test(inserting a bunch of rows and
arguably committing too frequently). Once I fixed that it was only 15%
slower which is negligible for OLTP.


--
Niall Litchfield
Oracle DBA
http://www.niall.litchfield.dial.pipex.com