Re: New architecture using Clusterware

From: Marko Sutic <>
Date: Thu, 17 Jan 2013 13:32:37 +0100
Message-ID: <>

Hello Dennis,
we had similar situation as yours with medium-sized databases on Solaris servers using VCS.

For new environment we created active/passive failover clusters using Oracle Clusterware on Linux with OCFS2 used for shared storage. (knocking on wood) We are running clusters now over a year without any problems.

Just after initial installation we had some storage problems cause of bad drivers and lousy FC cables/ports, but after we fixed that everything works without any errors.
Local disks are used for binaries and database files are on OCFS2 mounts. For managing resources we're using custom scripts.

I've heard lots of bad stuff about OCFS2 but I think the most problems happen if you use it to store large number of files. It is not very flexible system as ASM and maybe it has slightly poorer performance then ASM but for us it serves its purpose.

All my doubts were pointed to OCFS2 and I've spent over a month performing various stress/load/failover tests.
OCFS2 successfully survived all of them.

Are there better solutions then this, probably, but even on the most expensive solutions you could expect bugs and problems. Biggest benefit of this solution is cost-effectiveness.

I hope everything will work fine in next years as it worked till now :-)


On Tue, Jan 15, 2013 at 10:28 PM, Dennis Williams <> wrote:

> List,
> Traditionally we've supported databases with Solaris servers using VCS.
> These are medium-sized databases with average availability requirements.
> Nothing leading-edge.
> We are considering a new cluster of servers. I'm wondering if Linux and
> Oracle Clusterware (but not RAC) is a cost-effective solution that would
> provide adequate availability. Has anyone on this list taken that approach?
> Thanks,
> Dennis Williams
> --

*Marko Sutic*
My LinkedIn Profile <>

Received on Thu Jan 17 2013 - 13:32:37 CET

Original text of this message