TSM Server migration from HP blade to UCS

[rpandey]

Hi
I am planning to migrate Linux TSM server from HP blade to to UCS. Server physical, running TSM v7.1 on HP physical linux
machine. Could you please help to provide detail process on how can i move TSM server from HP blade to UCS?

 

[marclant]

There's multiple methods to migrate:

Option 1 (if retention period is short)
- build new server
- switch client to new server and start backing up
- decommission the old server after the retention period has passed

Option 2 (need same OS type and same Spectrum Protect version)
- backup db on old system, shut it down
- swing storage to new system, make sure all disk and file storage pools have the same mountpoints
- restore DB on new system and bring up the server
- if also using tape, may need to adjust the paths as the device names might be different.

Option 3 (if also moving to new storage)
- build new server
- setup server to server replication between both servers: https://www.ibm.com/support/knowled...bm.itsm.srv.doc/t_repl_default_configure.html
- replicate the nodes from the old server to the new server: https://www.ibm.com/support/knowled...m.itsm.srv.doc/t_repl_command_processing.html
- on the new server, remove the node replication: https://www.ibm.com/support/knowled...itsm.srv.doc/t_repl_configuration_remove.html
- switch nodes to new server
- decommission the old server

 

[rpandey]

Thanks for the quick response.
Can't i physically move Linux physical server to UCS?

 

[marclant]

That's not moving Spectrum Protect, that's moving the entire server, operating system and all. That would be more of a Linux question than a Spectrum Protect question.

Keep the following in mind those if you plan to move Spectrum Protect from physical to virtual:

All IBM Spectrum Protect products and components are supported on physical machines that have been physically or virtually partitioned such that each partition is provisioned with dedicated, discrete resources (CPU, memory, I/O).

That means not sharing CPU, NIC, HBA, memory with other virtual machines.

• IBM can make no guarantees with respect to the performance and scalability in a virtualized environment
• No support for attached tape drivers or tape libraries, either virtual or physically attached
• No support for LAN-Free data movement to tape or disk

source: https://www.ibm.com/support/pages/node/83755

All disk and file storage pools will need to have the same mount points in order to access them as well.

 

[Trident]

Hi,

I have used replication method lately for migrating from server to another, but only when data is on disk only. For larger systems, where data is on tape, I create new nodes on new system, and let the old data expire. When retention is done, delete nodes. For the ones where you have longer retention, like years, ren nodes to something _old, and export/replicate those to the new system. You need to concider the amount of data you are going to move/replicate/export. Do you have enough tapedrives to handle daily work and data replication?

Good luck,