TSM virtual volumes for export/import nodes data

[sra006]

Hi ,
I am evaluating using virtual volumes to store exported data from tsm clients.
I have time to export nodes from source to virtual volumes, and at some point I would need to import them to
the target to be nodes on the target.
My question is :
will it improve import time if I use virtual volumes that already on the target server to create source nodes on the target? vs directly using server to server communications for exports ?
Virtual volumes are controlled by the source server, does it mean that all I/o will be looping back to the source,
even if it resides on the target?
Thank you,
Al

 

[marclant]

will it improve import time if I use virtual volumes that already on the target server to create source nodes on the target? vs directly using server to server communications for exports ?

No because you would be moving the data twice.

Virtual volumes are controlled by the source server, does it mean that all I/o will be looping back to the source,
even if it resides on the target?

The I/O is handled by both the source and target. Source is doing I/O over the network to the target, the target is doing the physical I/O to the media (tape or disk).

 

[marclant]

I have time to export nodes from source to virtual volumes, and at some point I would need to import them to

Have you considered replication for this?

 

[sra006]

Thank you for the replies.
One of the reasons I could not run exports or replication is because my final target server does not exist ,
it will be build in 2 months. Meanwhile I was considering other options, that will avoid creating multiple tsm nodes on the "in-between target server"
As I see in one of the replies the target server will control the virtual volume pools for staging data .
My process for migrating a node :
source_srv ---> export to virtual v on "temp target" pool
import from "temp target" to the final target

In the scenario of exporting a node to a virtual volume on the "temp" target and importing it to the final target
will the data flow directly to the final target from the "virtual volumes" and just metadata gets referenced by the source?
This is what I am trying to clarify.

Best ,
Al