Hello there,
I'm using TSM 5.1 and and I'm trying to create a redundant TSM setup
I've a rather specific question on volume access in case of disaster tests. We're currently setting up a second library on another site . The idea is that the first library contains all primary pools and the second (offsite) library contains the copy pools. In the next phase we'll also cluster the TSM server . So that we got almost all aspects of DRP/BCP covered.
But there's one situation were I'm not sure what TSM will do. Let's suppose we lose access to the primary library for a few days because of a HW failure, so all primary tapepools are unavailable .
In the meanwhile we still want to take backups, so we assign some scratch volumes to a primary tapepool in the second library and we're ready to go. Afterwards we just need to move those volumes back to the first library and the primary pool is consistent .
But what happens if we want to do a restore. The primary pool is unavailable and as far as I know, TSM will only access the copypools if I set my primary volumes to access 'DESTROYED' . So at this point I'm able to restore, but how can I recover my primary pool after the HW failure is solved . You can of course start a restore storagepool (this would take days to restore all the data) or can you just change the access of the primary volumes back to readwrite .
Has anyone already tested this ?
What will happen with the incremental backups that are taken during the time that the volumes are in DESTROYED mode. Will TSM assume that he has lost all files and needs to take a complete backup instead of just an incremental ?
In my opinion TSM should be changed so that whenever a primary volume is unavailable it should try to access the data in the copypool and we should be able to change a copypool into a primary pool . If we have a real disaster and we lose our primary pool, I don't want to loose time restoring the copypool back to the primary pool.
I'm using TSM 5.1 and and I'm trying to create a redundant TSM setup
I've a rather specific question on volume access in case of disaster tests. We're currently setting up a second library on another site . The idea is that the first library contains all primary pools and the second (offsite) library contains the copy pools. In the next phase we'll also cluster the TSM server . So that we got almost all aspects of DRP/BCP covered.
But there's one situation were I'm not sure what TSM will do. Let's suppose we lose access to the primary library for a few days because of a HW failure, so all primary tapepools are unavailable .
In the meanwhile we still want to take backups, so we assign some scratch volumes to a primary tapepool in the second library and we're ready to go. Afterwards we just need to move those volumes back to the first library and the primary pool is consistent .
But what happens if we want to do a restore. The primary pool is unavailable and as far as I know, TSM will only access the copypools if I set my primary volumes to access 'DESTROYED' . So at this point I'm able to restore, but how can I recover my primary pool after the HW failure is solved . You can of course start a restore storagepool (this would take days to restore all the data) or can you just change the access of the primary volumes back to readwrite .
Has anyone already tested this ?
What will happen with the incremental backups that are taken during the time that the volumes are in DESTROYED mode. Will TSM assume that he has lost all files and needs to take a complete backup instead of just an incremental ?
In my opinion TSM should be changed so that whenever a primary volume is unavailable it should try to access the data in the copypool and we should be able to change a copypool into a primary pool . If we have a real disaster and we lose our primary pool, I don't want to loose time restoring the copypool back to the primary pool.
