Metrocluster with Continuous Access XP - Oracle RAC with ASM -- HP Unix

[ManojK_1]

Hi,

If anybody experienced Metrocluster with Oracle RAC with ASM (RAW Disks) Please share the knowledge.

Currently my production environment is working with 2 Node HP Unix 11.31 + SG + SGeRAC + HP Storage Works XP24000 storage.

We are planning to configure metrocluster with the above environment with a DR environment having same configuration. How can i configure this environment?

Also please share the Metrocluster package Failover and Fallback technique?


Manoj K

[INH]

Manoj,

this may give some information for your query

http://h71028.www7.hp.com/enterprise/w1/en/os/hpux11i-serviceguard-metrocluster.html

Regards,
INH

[ManojK_1]

Hi,

HP already announced that Metrocluster is supported with Oracle RAC on ASM.

Regarding this i have gone through so many documents, but no where it is mentioned about raw disks. Everywhere it is about Oracle ASM on CFS,LVM, SLVM, VxVM & CVM.

ASM on raw disks in Metro cluster is supported?

Manoj K

[Alzhy]

Well if you already have Continous Access betwwen your XP24000@SiteA and XP24000@SiteB - what do you specifically NEED?

Do you need procedures on how to establish your environment at SiteB in the event of a total outahe of SiteA and vice versa?

Let us know.

Also -- how are your ASM Disks configured in your existing 2 node cluster? Are you using ASM straight on multipathed disks OR ASM over RAW VxVM or LVM logical volumes?

[ManojK_1]

Hi Alzy,

I am using ASM straight on multipathed disks.
Is this supported by HP Metro Cluster?

I have gone through so many documents and every where it is mentioned as Metro cluster is supported with ORACLE RAC with ASM over CFS, RAW VxVM or LVM logical volumes.

So i am just confused and wanted to know the support of MetroCluster with ORACLE RAC with ASM on multipathed disks?

Manoj K

[Alzhy]

Well in your case - if you're using ASM direct on top of /dev/rdsk/diskNNN devices - then I think it is easier.

As a backgrounder:

ASM Instantiates its storage these way:

- +ASM DB Instance (the ASM Engine) starts
- checks for all disks in "asm_disksrting" for "ASM signatures"
- figures out the Diskgroups from among the disks
- "mounts" the diskgroups and redies them for service to the Databases

Since you're not using a VOlume Manager - then as long as the member disks are complete on the other nodes - then you should have no problem. The disks do not necessarily even need to have the same name or cXtYdz on the other nodes as +ASM scans disks based on "asm_diskstring" and the disk "signatures" on each.