Backup/Restore approach when installing new RAID controller

[BoyeDav]

We're running OpenVMS 8.3 on an Integrity rx3600.

We have (8) 73GB SAS disks connected using the standard 8 port SAS HBA (347786-B21) that shipped with the rx3600. We're upgrading to the SmartArray P400 SAS controller because of hardware support for the RAID levels we want and because it has a battery-backed cache.

Our current system volume is two drives using hardware RAID 1. Our other six drives are software RAID 1+0 using RAIDSERVER with shadowed drives. We use that for our data volumes, and we have multiple volumes created using initialize/size=.

The new controller and disks will be setup using hardware RAID 10. I'm guessing OpenVMS will see this as a single device, and we're going to create multiple logical volumes.

I'm looking at the following as a reference for doing the backup/restore:

http://h71000.www7.hp.com/doc/83final/ba322_90045/ape.html

However, I'd appreciate any other guidance you might have for this type of situation.

Thanks!

[Hoff · Accepted solution]

ECOs to current. Initialize the target disk with a cluster size to a multiple of 16 and more than enough headers, and follow the directions for a disk-to-disk BACKUP /IMAGE /NOINIT transfer. Enable Dynamic Volume Expansion.

Here is what I have been classically recommending for the disk initialization process:

http://64.223.189.234/node/193

For the paranoid and given measured disk failure rates, I'd also redistribute the disks in the array, and scatter them across multiple controllers; I'd ensure the disks in the array are from different batches and different sources. (RAID5 has a nasty habit of encountering a secondary spindle failure during the recovery from a failed spindle, and tipping over hard, for instance.)

If you are unfamiliar or uncomfortable with OpenVMS system operations and/or BACKUP and/or sufficiently paranoid over your data, do consider enlisting formal assistance.

Stephen Hoffman
HoffmanLabs LLC