Slowness with HA targets mounted

oxyi · Wed Mar 07, 2012 5:18 pm

Yes, I've changed the traffic priority long time ago to Faster synchronization, is that correct ?

Yes, no IOMeter yet, I can do that and post back.

But for my question that still hasn't answered yet, if I don't mount the two HA targets, will the data still by sync or not ? It should right ?

Fri Mar 09, 2012 10:36 am

I would actually try to move crossfader to 50% first and after to "Client Requests". Could you please try that?

And yes, if the sync was initiated it will continue even if client machines will be disconnected from targets.

oxyi · Wed Mar 28, 2012 10:22 pm

Sorry for the delayed reply.

The sync channel between the two storage unit is at

827Mbits/sec
833Mbits/sec
831Mbits/sec
829Mbits/sec

Not the best, but decent enough to transport data at least 60 MB/sec.

It was actually at 50%. Then since I was having issue, I was advise to move to faster synchronization, but it is still the same.

Thu Mar 29, 2012 2:30 pm

OK, I`m a little bit confused now.
Are you issuing slow synchronization speed or slow client requests processing?

oxyi · Wed Apr 04, 2012 8:43 pm

Hi Anatoly,

Before it was set at 100% over at Syncho speed, and now I've changed to 50% fade, and the speed was still same.

Are you asking me to move it all the way to client request ?

Thu Apr 05, 2012 10:29 am

Dear oxyi,

I am asking during what exactly process and at what data link are you experiencing the performance issue?

Thank you

oxyi · Thu Apr 05, 2012 5:37 pm

I am experiencing the performance issue as soon as I attached my HAPartner, so when you are asking which datalink, it doesn't applied to me.

Each datalink are benchmarked and result were all goods, as soon as I connect my HAPartner, my performance just dropped, that's what I want to figure out.

Fri Apr 06, 2012 9:24 am

Well, as far as I remember your SyncChannel data link was using switch and VLAN, and I mentioned that we strongly recommend to conenect servers directly for this purposes. Have you accomplished this please?

oxyi · Tue Apr 10, 2012 4:12 pm

Yes, I remembered you guys told me it is not great, but I do not have a choice, and I am pretty sure there are customer out there that are doing it via switch as well.

To make it simpler, I put the synch channel on a Cisco 2960 switch all by itself, and benchmark it, and speed was excellent.

If my benchmark is good, why does it matter if it directly connected or not.

Thu Apr 12, 2012 2:52 pm

Dear oxui,

Would you kindly clarify what benchmarking through the switch have you just mentioned (configuration, specification, etc.)?

Thank you