|
The best way to do this with the most success and easiest failover is to use a cluster to share the disk resources between two nodes. You can use any available clustering technology you wish such as Microsoft Cluster Server (MSCS) or Veritas VCS. MSCS is included with the Enterprise and Data Center editions of Windows. Dynamic disks should not be used in a cluster.
I have had intermittent success without the cluster software installed, but you might try this technique: Make sure both servers are similar in their slot layout for the HBAs, and use BASIC disks when writing the disk signature on the first server. Never have both servers up at the same time when not in a cluster, or your data may become corrupt. Once the signature is written, export the DISK registry subkey parameters for the new disk, and import the subkey parameters into the other server. Back everything up before doing this, and make a backup of the registry keys on the server you are importing to in case anything goes wrong. You may not have to do the registry trick if the second server sees the disk and does not try to write a new signature.
A better approach would be to use SAN-based booting for the servers in question. Having your "C:" drive on the SAN allows a server of the same configuration to boot to the failed server's SAN-based root drive, and the new server would then acquire the exact same profile as the server that was lost. I call this technique "poor man's clustering." You need to have similar server hardware configurations for this to work properly. The good news is that I have a complete step-by-step guide on booting into SAN-based storage in chapter 10 of my book "Storage Area Networks for Dummies"(ISBN:0-7645-2480-1).
|
