Page 1 of 1

to big latency

Posted: Tue Sep 18, 2012 3:04 pm
by silverspy
Hi guys,
is this latency normal? physical is physical server and VV is thick. load is generated by restore. server dont have load.
vmware - vmware disk is thin and VV is thick. when VV is thin latency is around 30-40ms. load in graph is generated with svmotion. in datastore are 4 VMs with no load. thanks for opinion.
3par is V800 P10000 with 4 controllers, 1000HDDs, AO is off for NL disks. All data are in SSD or in FC tier.
i have same latency, in full loaded EVA8100.
thanks for help

Re: to big latency

Posted: Fri Sep 21, 2012 11:46 pm
by Richard Siemers
What latency? You're at less than 7 ms for the whole chart. I don't see any 40ms latency in either of those graphs.

In the top chart, I see 3 to 4 ms (not 30 to 40), with a peak of 7 ms. The chart can be confusing since its showing ms, iops and mb/s all on the same axis.

Can you clarify what your asking?

Thanks!

--Richard

Re: to big latency

Posted: Sun Sep 23, 2012 4:17 am
by Christian
Just a tip , run thin prov on the 3PAR and thick eager zeroed in VMware.

Re: to big latency

Posted: Tue Sep 25, 2012 1:07 am
by ilv
The only thing I notice , is that write time for the physical disk is bounded to 3 ms/IO while for vmware it is not bound to any value. I infer the first one is using synchronous replication and the second one is not . Am I right ?
Thank you
ilv.