|
Continued from Moving SCSI to Fibre - Part One
After the data has been copied to the FC Symmetrix disk volume completely, the SCSI Symmetrix disk volume can be taken off-line. In true Windows style you will probably need to reboot the system.
The "Still pretty lucky" case
A weird corner case perhaps, but could apply in some scenarios. Assumes that you have time and can halt user access and application processing, but there are no slots available for the FC adapter. There are two potential transfer methods in this case: backup and network copy. If you use backup, you will need to restore, and that means you will need a spare system with a FC adapter to connect to the FC Symmetrix disk volume to restore the data through. After the restore completes, the production server needs to exchange a SCSI adapter for an FC adapter and connect to the FC Symmetrix disk volume.
The "To be expected case"
Assumes you don't have time to quiet the system to get a good copy of all the data. Mirroring is probably the best approach here, but this assumes there is room for an FC adapter in the production system. The downside of this is that you can't expand the size of the FC Symmetrix disk volume. The mirror will have to be a software mirror, which can be done in either NT or Win2000. At some time in the future it may be possible to make the FC disk volume larger by extending it with the NT disk management tool. Again, an experienced NT administrator is likely to be more familiar with this capability then I am.
At some point, the mirror will need to be broken to remove the SCSI Symmetrix disk volume. This is much easier if the operating system's master boot record does not have to be created on the FC disk volume.
The "I've got the money" case
An alternative to mirroring is to use a replication product, such as the Veritas Storage replicator. This involves the use of a spare server that connects to the FC Symmetrix disk volumes to replicate data to. A single spare server can manage the replication of all the production systems, but I wouldn't recommend overloading the spare server this way. You may be able to talk to Veritas and get some idea on the ratio to use. I haven't used this product myself, but I have a fairly high regard for Veritas as a software company. There is something to be said for automation managing this type of process.
To summarize the big picture:
* Take the OS boot record out of the equation by migrating the OS to another local storage disk(s)
* Establish I/O path segregation between servers and FC
* Attach the FC disk volume to the server or to a spare server
* Transfer all the data from SCSI to FC disk volumes, expect to quiesce the server to get a complete image
* Remove the SCSI volume and add the FC volume to the server if it is not yet
Hope this helps in some way.
|
