Re: ASM Question

From: jason arneil <>
Date: Fri, 5 Dec 2008 20:47:21 +0000
Message-ID: <>

2008/12/5 Sebastião Carlos <>

> To understanding my question see the details of the environment:
> Production Site:
> - Database Server IBM - P590 - AIX 5.3
> - Storage IBM DS4800
> - Filesystem - JFS2
> - Cluster IBM HACMP
> Standby Site:
> - Database Server IBM - P570 - AIX 5.3
> - Storage IBM DS4700
> - Filesystem - JFS2
> - Cluster IBM HACMP
> The applications (ERP, CRM, etc.) access the instances of database on
> Production Site.
> If I have any fault on hardware/software of P590 or in DS4800 storage the
> HACMP migrates
> all environment of database for the Standby Site. Logically in this
> scenario, case occurs
> a failure on hardware/software, will unavailability of database for a few
> minutes until what
> this instance is started in Standby Site.
> For this to work exists between the filesystem used for instance a
> mechanism to mirror
> designed in AIX/HACMP that replicates the modified data in filesystems of
> DS4800 to DS4700.
> Of course, the same filesystem structure exists in a production site and
> standby site.
> My big problem is that this solution is not able to replicate raw devices.
> I plan to put the instance in ASM. In a short period of time want to use
> RAC to use the ability
> processing of the P570.

This is an extended RAC setup. You need to be aware ASM will write primary extents to both your failure groups. Mirroring in ASM is not like traditional hardware RAID mirroring. In 10g, ASM always trys to read primary extents. So if this is what you are running, your application will need to read the data from your standby site. You will need good connectivity between the sites.

This is improved in 11g with the ASM_PREFERRED_READ_FAILURE_GROUPS parameter meaning you can direct an instance to read from just 1 side of the mirror.

> Now my question is: If my ASM diskgroups are created in the DS4800 with
> normal
> reduncandy and the fail group is set for the DS4700 at the time of
> unavailability
> of the storage DS4800 I could start the instance of database from fail
> group in
> DS4700 storage?
> To recover the DS4800 could migrate database to DS4700 from DS4800?



> --
> Sebastião Carlos Santos
> Oracle Database Administrator - DBA
> Oracle Certified Professional - OCP 8i, 9i e 10g
Received on Fri Dec 05 2008 - 14:47:21 CST

Original text of this message