Hardware RAID for multiple volumes within individual SAN in P4500 G2 Cluster

[epols]

So I've logged in with the CMC, and I've set both of my SANs (7.2TB each, 14.4 clustered) in the same management group. I've configured each one to be RAID 1+0.

My question is.. Can I not setup volumes across HW raid within each 7.2TB SAN, or can it only be done with network RAID across the two 7.2 SANs? Given that they're two physically separate machines, I had assumed I could configure volumes (with the same raid configuration) across multiple disks within each SAN. I'm planning on having an oracle database on a 2.4 TB volume, and I wanted to sort of isolate that on one of SANs and use the rest of the storage as a place for just vm image backups and nothing else I/O intensive. It appears that the only options I have when configuring a storage pool are to change the RAID level of the entire device, but not to assign individual disks and create volumes. Am I missing something?

Thanks!

[Fred Blum]

You have to differeniate between HD raid and network Raid. HD Raid is set per node, not per disk within the node. I agree that that would be a desirable feature. nRaid is configurable per volume.
I went for HD Raid 10 and nRaid 10 turning a 7.2TB P4300 starter SAN into a 1.8 TB solution.

Performance and High Availabilty wise why would you want to isolate the oracle database on one node?

When nics are bonded ALB you get transmit 2GB teamed but still 1GB receive. Or 2GB TX/RX with LACP when your switch config supports it.

When Virtual IP and Load Balancing is enabled within a cluster the load is distributed across the nodes within the cluster. So you get more NICs, more spindles and providing HA.

[teledata]

The only way to isolate storage to particular P4000 storage node is to create a cluster of 1 node.

The act of creating a cluster makes the following statement true:

"ANY Volume created (regardless of your network RAID selection) will ALWAYS be striped across EVERY disk and EVERY node within your cluster."

[epols]

So I'd like to achieve RAID 1+0, turning 14.4 TB into 7.2 usable. Any recommendations on the configuration to do so? HW Raid 1 on each SAN, nRAID 0 between them?

[teledata · Accepted solution]

If you used hardware RAID 10 AND wanted to maintain high availability with NW RAID 10 (volume stays up if a single node fails) your effective usable would only be 3.21TB. Unless you have VERY high Write IO requirements this configuration would be overkill and inefficient.

I recommend each node configured with Hardware RAID 5 (actually with P4500 it is a hybrid of RAID5+0 or RAID 50)

Then Network RAID 10 (2-way Mirror) for your critical volumes:
If you did this, and did NW RAID 10 for all your volumes you would have an effective usable 5.35TB after hardware and Network RAID overhead. If you have some volumes where high availability it not important (temp/staging/test etc) you could use Network RAID 0 and increase your capacity.

Each node in the above configuration would be 7.2TB (advertised), 6.42TB (right sized), 5.35TB (after Hardware RAID 50)

[epols]

Thank you for the explanation. I assigned you points.