This post is more than 5 years old
21 Posts
0
587
How can I pull back my EFD allocated through FASTVP?
We recently migrated over to a VMAX 10K, and turned on FASTVP for our Primary Healthcare application here. It quickly took over all of the EFD disk that FAST had allocated to other applications. Unfortunately, due to the way the application stores and accesses data, as well as the typical life cycle of Patient data in a hospital, FAST doesn't seem to have made any noticeable change in performance.
So, I'd like to pull back the EFD disk that has been allocated to this Storage Group. I was told that if I just disassociate the Storage Group with the FAST policy, the data that's currently on EFD disk, will just stay on the EFD disk. Is that correct? If so, how can I get all of the data from the Storage Group back to just FC disk?
Thank you
SYMCLIGuy
98 Posts
1
October 17th, 2013 09:00
Dynamox,
yes, if you changed the policy the SG was associated with to exclude EFD, FAST VP would slowly move data out of EFD to become compliant with the policy. The keyword here, though, is slowly. And I mean slooooooooooowly.
Compliance movement requests generated by FAST VP are limited to a maximum of 10GB, based on an FRR of 1. If the FRR is something other than 1 then divide 10GB by that number to get the new max size of a compliance request. eg at the default FRR of 5, the max compliance request size is 2GB.
Compliance requests are only generated approximately once every 10-15 minutes. If you're only moving 2GB every 10 minutes it's going to take a while to move, potentially, TBs of data.
The VLUN migration Paul suggested is the best and fastest course of action in this case. However, the -src_pool option requires the array to be running 5876 code, so hopefully that is the case.
PS the max movement request size tied to the FRR only applies to compliance movements, it does not apply to performance related movements.
dynamox
1 Rookie
1 Rookie
•
20.4K Posts
0
October 17th, 2013 07:00
Paul,
how about creating new policy that only includes FC drives, would it force "out of compliance" and move all extents back to FC tier ?
rawstorage
419 Posts
0
October 17th, 2013 07:00
You can do a vlun migration only selecting the EFD tier so only allocations from there will get moved
symmigrate -sid xxx -sg groupname -src_pool EFD -tgt_pool FC
you might want to adjust the FAST policy to reduce the EFD consumption too
dynamox
1 Rookie
1 Rookie
•
20.4K Posts
0
October 17th, 2013 10:00
but based on what you explained above i am actually getting an SRR (slow relocation rate)
SYMCLIGuy
98 Posts
0
October 17th, 2013 10:00
The FRR is the FAST Relocation Rate - it's a QoS (quality of service) setting for FAST VP.
dynamox
1 Rookie
1 Rookie
•
20.4K Posts
0
October 17th, 2013 10:00
make sense, thank you . What does FRR stand for ?
Barnzy
21 Posts
0
October 17th, 2013 12:00
Thanks for the info guys. I think that due to the nature of this data, we may opt for the Slow move back. It's only about 1.5TB. I don't have an urgent need to have it off of the EFD disk, I would just like to make it available to other applications again at some point.