Strategy for Data Migration from Veritas Cluster to Solaris Cluster
By hhnguyen on Sep 12, 2007
Heres is the second part on migrating from Veritas Cluster to Solaris Cluster as promised.
When you migrate from Veritas Cluster Server to Solaris Cluster, you'll also have the opportunity to cut cost even more by switching its underlying volume manager and its file system to one of the free options built in Solaris, like Solaris Volume Manager and ZFS. The choice is yours if you would like to keep the Veritas Volume Manager and the Veritas File System.
You should have a data migration plan developed as soon as you think about migrating an existing cluster. If you create a completely new cluster with new data, there is no need for migration plan obviously.
In principle, two strategies can be employed to move your existing data from a Veritas Cluster Server to a Solaris Cluster
1. Just use the storage with Solaris Cluster. You can go down this path only when the volume manager and the file system are unchanged, and the underlying storage is supported with Solaris Cluster. During the testing phases of the migration, you can very easily work with a backup of your data.
It is possible to migrate an cluster in place by splitting it into two parts. You would need to allocate a fraction of your server and storage redundancy during the migration, so the migration could be carried out as quickly as possible.
2. Take a backup of your data and restore it on a Solaris Cluster. This has to be done whenever the volume manager and/or the file system are changed. For example, if you change from Veritas Volume Manager and Veritas File System to ZFS, it will be a backup-and-restore exercise.
Splitting up a cluster for an in-place migration
Let us assume a two-node cluster where you want to migrate from Veritas Cluster Server with Veritas Volume Manager to Solaris Cluster with Solaris Volume Manager. Your data are presumably mirrored. Below are the main steps of the migration process. To get the necessary cluster administration commands, please consult with the applicable documentation.
1. Reduce the Veritas Cluster Server to an one-node cluster, and disconnect the interconnect. The interconnect has to be disconnected so that Solaris Cluster can be installed on the other node. Solaris Cluster does check the interconnect for unwanted traffic.
2. Split the storage in two halves, and disable the access from the Veritas Cluster Server nodes to the to-be-installed Solaris Cluster. This can be accomplished, for example, by modifying the switch zoning or lun masking. At this point, your application is still running, but you temporarily have no high availability and no data redundancy.
3. Install a single node Solaris Cluster on the second host. A fresh Solaris install is highly recommended.
4. Configure the full Solaris Cluster topology with a temporary copy of your data. Because the volume manager is changed, data will have to be installed via backup/restore procedures. It is important that you use different IP addresses for the logical hosts to avoid duplicated addresses. The new single node Solaris Cluster is now ready to take on the actual data.
5. When you are ready for an application downtime, transfer the actual data from the Veritas Cluster Server again to the Solaris Cluster, and shut down the remaining Veritas Cluster Server single node cluster.
6. Change the IP addresses of the logical host in the Solaris Cluster to the final values and enable all relevant resources. Your application will run on the new Solaris Cluster from now on.
7. Reestablish the interconnect, destroy the Veritas Cluster Server and install Solaris Cluster packages on the old Veritas Cluster Server node, but do not configure the node yet.
8. Enable data access to the storage for both nodes using appropriate methods.
9. Add the second node to the Solaris Cluster including the Solaris Cluster device groups. This step will take another brief application downtime.
10. Mirror your data. You now have full redundancy and full high availability once again.
Congratulations, your migration from Veritas Cluster to Solaris Cluster has just been successfully completed!
As previously stated, it is much easier to migrate on a hardware refresh, but if you can tolerate a temporarily compromised availability during the migration process, it is very feasible to migrate from Veritas Cluster Server to Solaris Cluster in place.
Sun Cluster Engineering