Grigor,
With large sata/fata drives recover can take a long time and during that time
one could loose a second disk drive which with RAID 5 would result in total
data loss.
RaID6 provides two sets of independent party calculations P is one set and Q
the other so you can have dual disk failure
Grigori,
Regarding the point made below:
>>1) migration is totally controlled by TSM Server and depends on number of
>>nodes and number of migration processes. I am afraid to overlap migration
>>activity and backup/restore operations.<<
I don't know how big your disk pool is but you should be a
RAID6 came together with FATA 900GB disks.
IBM does not support RAID5 on so big disks (only RAID10 and new RAID6).
Difference in arrays is: RAID5 - 6+P+S, RAID6 - 5+P+Q+S.
I do not know what is drive Q, but it allows to recover big disks faster.
Grigori G. Solonovitch
Senior Technical Architect
Hello Grigori
Quite new then if you're using raid 6.
In that case, I'd not worry too much about fragmentation. If you have
more than a couple arrays for each of the FATA drives, the data is
probably going to be all over the drives anyway.
steven
Steven Langdale
Global Information Services
EA
Hello Steven,
We have IBM DS8100 with:
- 146GB Enterprise disks (RAID5) for data;
- 500GB FATA (RAID5) for backup and archive data;
- 900GB FATA (RAID6) for backup and archive data.
DS8100 is shared by nodes and TSM Server using different arrays for data and
backup data.
Regards,
Grigori G.
Hello Grigori
What SAN have you got, and is it shared storage? I ask because there has
been discussion on FS fragmentation but surely this is meaningless if your
using a SAN.
Steven Langdale
Global Information Services
EAME SAN/Storage Planning and Implementation
( Phone : +44 (0)1733 584175
(
Hello Christian,
Thank you very much for your proposal. I understand that it is much easier to
use migration, but I am still going to use procedure based on data movement
because of:
1) migration is totally controlled by TSM Server and depends on number of
nodes and number of migration pro
Grigori,
Your process below works; I would have done the same as Christian
Svensson that is migrating all the data off of the Disk storage pool to
the next storage pool (what hierarchy do you have in your environment:
Disk -> VTL -> Tape or Disk -> Tape?? ). Depending on the number of
tape drives
