26.3 Migration Options

How then do you migrate a V1.[56] cluster to V5.X when you can't perform a rolling upgrade and you can't have the V5.X cluster connected to the V1.[56] cluster (in any way except possibly on a LAN)? You have three options from which to choose. These options are usually thought of as a trade-off between downtime and expense.

Option 1:

Create a separate cluster consisting of new systems and storage. Once this cluster is built and tested, migrate the functionality from the V1.[56] cluster to this new V5.X cluster. The problem is that few people have the luxury of obtaining this additional hardware.
Option 2:

Create a separate, perhaps limited cluster with new systems and at least enough storage to establish the cluster and test your applications. Once the new cluster is fully configured and tested, physically connect the old storage from the V1.[56] cluster to the new cluster. This option is slightly less expensive than option 1 but still requires additional hardware.
Option 3:

Upgrade the existing V1.[56] cluster using its hardware and storage. Remove and disconnect one member from the V1.[56] cluster, install and configure a standalone Tru64 UNIX V5.X system, shut down the remaining systems in the V1.[56] cluster, connect the V1.[56]'s storage to the standalone V5.X system, configure the storage, create a single-member cluster, and then add the other systems to the new cluster. This option requires the least amount of additional hardware (perhaps some additional disks, HBAs, and LAN interconnect cards (if using LAN interconnects for the cluster interconnect)).