The Missing Links To The Resources You Need!
There are a number of words you'll read before becoming uninterested. You can read everything below in under 20 seconds.Based on real word advantages:
TSM Disk does fragment since all data is stored in aggregates. If you decide to store data long term in disk pools for any reason, you will actually end up wasting space as data is expired and the aggregates become fragmented. Without the ability to reconstruct the aggregates and reclaim the empty space in them you will need to perform MOVE DATA commands to alternate volumes with the RECONSTRUCT option set to yes to get the space back. This is listed in the list given here:http://publib.boulder.ibm.com/infocenter/tivihelp/v1r1/topic/com.ibm.itsmaixn.doc/anragd5583.htm#wq137You'll notice it states that aggregate reconstruction is not available for disk volumes and the result is wasted space.
I believe that you can only do single session restores from Disk, rather than a multi-session restore from File. I think the earlier post regarding fragementation was with regard to file system fragementation. Provided that the disk pool volumes are created serially on the filesystem, then they should never fragement at a file system level. On the other hand files created using the file dev class are typically created and deleted dynamically that eventually causes some fragmentation at the file system level. What I have not ascertained is what the true performance impact of this is in the real world. It's true that disk pool volumes cannot be reclaimed and that theoretically this could lead to waste space. In reality, I've not found that there is not that much space wasted due to this occurrence... at least in the instances where I've measured it.. it's been about 2-4% of the total disk pool. Another interesting thing is that if you use space triggers to create file volumes dynamically, that TSM doesn't actually delete the physical file if the logical volume is deleted - you have to do that manually. LAN-free to disk? Using SANergy? I think I just threw up in my mouth.
Thanks for the comments guys
CHADSo MOVE DATA + RECONTRUCT would resolve the issue on DISK pools or there is no option to resolve fragmentation?
Unless you decided on using a disk pool for long term storage then you wont have any issues with wasted space. If you do plan on using a disk pool for long term storage then the only way to reclaim empty space in aggregates is the MOVE DATA with the RECONSTRUCT=YES parameter.