3par 8200 bad performance (sequential write hangs up the storage)

[OAndras]

hello all,

we have a little san environment: 3par 8200 + 2 san sw (brocade 8/24) + servers:

3 x hp dl380 gen9 - wmware

2 x hp dl380 gen8 - HyperV

2 x 2 ibm 3620m4 oracle rac

the 3par 8200 has 38 x 1.8Tb SAS 10k SFF HDD (FC) + 10 x 1.92TB SAS SFF SSD (SSD).

with normal load everything looks good, but: when theres is a bigger data copy form FC vvol to FC vvol - all FC vvol are hanging.

So there is a data migration from oracle asm from FC (CPG: RAID6 - 38 x FC disk) to an other FC vvol (same CGP),

the data move speed is looks normal (800.000 - 1.000.000 KBs/sec - 3par_perf_sys_tp.jpg), but all other systems (vmware, hyperv) hangs... In vmware the latency 250-500 ms!!! (3par_perf_vmware_latency-avarage.jpg) the normal latency 0-15 ms!

I made a test: there are 3 vmware vms (2 located in FC vol, and one in SSD vol), i copied bigger file (via nfs):

in normal load: FC to FC 10 sec, FC to SSD 8 sec

while te data move is in progress: FC to FC 40 min!!!! 40 _min_ not sec!!! FC to SSD 28 sec

so I think all the write performance is reserved the sequential write of data migrating... the read performance is slower but maybe it is normal while heavy load (FC to SSD copy 26 sec, so the read performace is slower than the normal, but not 100x times)

this very poor performance / paralell working is normal? what can we do? we have to move and migration data within th FC quite often! Is the QoS help? We haven't got licence for QoS, so unfortunately we cannot try it...

i attached 3 pics from latency and throughput

thank you!

Andras