This post is more than 5 years old
50 Posts
0
9054
Symmetrix VMAX performance troubleshooting
Hi all,
I wanted to ask a general question for Symmetrix VMAX performance troubleshooting :
1) What to measure generally for performance issues ?
2) Any one has any notes that explains the Metrics used in SPA.
3) If I have a LUN bound to a pool, how can I bind it to another pool without affecting access to it or data loss.
4) Does FAST affects any performance for any Databases etc ????
Any thing you can share on performance issues for VMAX.
Regards,
Kavan
AJK3
108 Posts
0
March 9th, 2012 07:00
Good question.
swarup_kundu
5 Posts
0
March 9th, 2012 07:00
Hi Santosh
I have just one confusion.Since ,here the source lun is concatenated...the BCV device would be stripped one or concat?
sracharla
49 Posts
1
March 9th, 2012 07:00
Swarup,
If the conerting from a concat meta to stripe meta, i.e., if the source is concatenated, then the BCV meta needs to be concatenated with size and member count equal to the source concatenated meta.
If you are just expanding a stripe meta by adding more meta members, the BCV meta used should be striped meta with size, members and stripe size equal to that of the source striped meta.
swarup_kundu
5 Posts
0
March 9th, 2012 08:00
Thanks Santosh.
Kavan2
50 Posts
0
March 12th, 2012 05:00
Rebind of thin devs from SATA -> FC.
I know this can be done using rebind where in all new IO will be on FC and existing data will still be on SATA.
but is there a way to move existiing data on FC as well??
During this operation I take it LUN can be still online to applicatons .
Kavan2
50 Posts
0
March 13th, 2012 03:00
In one of my original questin... for FAST...
during the data migration.. is it good to enabel/use FAST ???? or is it best to leave it till data migration is complete ??
dynamox
1 Rookie
1 Rookie
•
20.4K Posts
1
March 13th, 2012 05:00
when my DBAs are using Oracle ASM to migrate data from DMX3 to VMAX, i do not enable FAST on the storage group until after they complete their migration. I figured FAST would be trashing back and forth during the migration so i prefer to let it "back-in" for a week and then enable FAST.
Kavan2
50 Posts
0
March 13th, 2012 09:00
Thanks Dynamox that’s helpful.
Kavan2
50 Posts
0
March 22nd, 2012 11:00
Hi guys,
back to the performance issue ,
i got a tdev : about 1.90TB meta of 240GB each meta meber. SATA drives.RAID6.
i grabed the stats from spa and i can see 5000,6000,2000 WP counts.
Is this normal, i think the figure is high. users are saying they are expereincing slowness etc. how can i decrease these.
dynamox
1 Rookie
1 Rookie
•
20.4K Posts
0
March 22nd, 2012 11:00
what is this device used for ?
Kavan2
50 Posts
0
March 22nd, 2012 12:00
Device is used for VMDK. Datastore.
dynamox
1 Rookie
1 Rookie
•
20.4K Posts
0
March 22nd, 2012 12:00
probably not a very good i/o profile for raid-6, so many WP because SATA is slow, can't destage writes fast enough. Might need to move storage vmotion some VMs to other datastores or move this entire device to another tier.
Quincy561
1.3K Posts
0
March 22nd, 2012 13:00
5000 WP tracks may not be very high at all.
Check your device and system WP maximum with symcfg list -v
Kavan2
50 Posts
0
March 22nd, 2012 13:00
Thanks Quincy , that was what I was looking for, the theoretically possible value and the actual one.
Quincy561
1.3K Posts
0
March 23rd, 2012 05:00
What is the overall read rate? 200 may be a very small percentage.
Why concat? Do you need to expand regularly? Striped does have a performance advantage, while concat has an expansion advantage.