Yes you can do this with ASM setup the diskgroup as normal redundancy. Explain you would create two failgroups one call A and the other call B. In failgroup A you would put all the disk/LUNS from SAN A and in failgroup B you would put all disk/LUNS from SAN B.
We had the same issue with a SAN outage so we converted our ASM disk group to normal redundancy from cross cabinet outage issues.
If you send me your email I can send you a document I created during our testing so you can get more understand of the possible configuration.
1 person found this helpful
For ASM it does not matter what technology is behind your devices in terms of providing data redundancy. If you configure ASM with normal or high redundancy, ASM provides data redundancy between disk failure groups. ASM does however not mirror devices, and uses file extents (AU units) distributed on available free space of disk failure groups.
How you configure your ASM disk groups is up to you. By default, every device is its own failure group, but it it can be configured. If you have different physical paths to your devices, you might want to combine separate ASM disk groups by controller or storage paths for additional hardware fail-over.
ASM provides performance and redundancy from the point of data storage devices. It does not provide redundancy for your Oracle database instance or services. Oracle Dataguard provides service and data availability by the means of synchronizing data between a primary and standby database.
I sent you the documentation I created thanks.
1 person found this helpful
I've read some stuff on ASM but am not finding a clear answer, can I mirror all flow of data to our Oracle databases to the two different SAN's so that if SAN 1 breaks, SAN 2 takes over with no loss of service? Or would I still need some Dataguard or something to make this work?
I have a RAC that runs 2 storage arrays as mirrors via ASM.
Works great. Have had complete chassis failure of the one array, without the database being aware of it. Architecturally, whether it is an actual SAN, or a storage array, is immaterial to ASM. This particular I/O fabric layer is Infiniband and not Fibre Channel, but that does not impact ASM at all either the one way or the other.
Keys issues IMO are
- you still need full redundancy per SAN/storage array with multiple I/O paths via redundant swicthes
- you need similar I/O performance from both SANs/storage arrays
This was exactly what I was looking for.
I'm interested about the ASM and two SAN implementation.
Could I have your documentation?
Can I get it by FTP or dropbox?
Thanks a lot.
I still have it, I'll just need your email so I can send it.
I'm very interested in Javier's documentation, can I have it?