UNIX for Dummies Questions & Answers

I have heard that it is possible to have the following setup for running an Oracle database on HP-UX , but I would like a little confirmation.

Two servers, each running HP-UX 11, acting as a sort of tag-team partnership. They share SAN space for all their storage needs. One server, the primary, is running an Oracle database, but if it goes down for any reason, the second server can take over simply by downloading the binaries for Oracle from where they are stored on the SAN. Once these are downloaded (in a matter of minutes at the most), the second server can immediately start acting as the new primary. Prior to this download of compiled binaries, the second server had nothing but a few configuration files and scripts on it (no Oracle installation at all). I am assured that these pre-compiled binaries are all that are needed. No administrator interaction, all configurations have already been done.

Does this, in fact, sound like a reasonable and accurate thing or is someone trying to pull the wool over my eyes?

2 figures:

1) you have HP MC/ServiceGuard and so we are talking of a HA cluster, in which case if correctly configured it can to be up again in minutes or even less...

2) You are not in a HA cluster but you are on a SAN with 2 or more HBAs, in which case
you have to write the scripts to vgchange the volume groups of the defunct server you see the PV thanks to the SAN admins...after sanity check and mount adequately you ca start the new instance but you are stuck with 2 things:
a) the host's name
b) its IP address

It can work (I did it long ago with 2 D class...) lets say you have a developpement server and a production server almost identical, if the production server drops, you can sacrify your test server by having the adequate config scripts ready ( with passwd etc...) and after reboot it replaces the faulty prod, having same name and IP...

You are not limited to MC ServiceGuard. You can also use Veritas Cluster Server as long as you are also using Veritas Storage Foundation For Oracle to manage your storage. The most robust solution is to use Oracle RAC. This is a parallel implementation of Oracle, where all nodes share the work, and if one goes down there is NO INTERRUPTION.

It can be implemented as a straight Oracle solution, with Oracle managing the clustered file systems and cluster services, or it can be integrated with Veritas Storage Foundation for Oracle RAC, where the clustered volumes/filesystems are managed by Veritas CVM/CFS, and the cluster services are managed by Veritas Cluster Server.

I would not rely on a homegrown solution, unless the budget isn't there for RAC/HA.