SCVMM live migrations very slow

Question

I'm running Hyper V with SCVMM 2012 in my test environment. 2 hosts connected to a MD1200 and set up in a cluster. Everything works perfectly, except that live migrations take a super long time. By this I mean that when I start the Migration Wizard it takes up to 5 minutes to rate the hosts, and then when it finally completes and I start the migration, it can take up to another 5 minutes to complete the move.In my production clusters this doesn't work this way, there the Migration Wizard takes seconds to rate the hosts and usually less than a minute to move the VM.The only major difference between my test environment and my production environment is that test is using direct SAS connections to the SAN, where in prod everything runs through switches. In prod one cluster is all 1gig and one cluster is 10gig and both are about the same...
This topic first appeared in the Spiceworks Community

Answer 1

Hey guys,
I have searched online and in MS forums, and cannot resolve my issue.
I have checked through similar articles, without any resolution.
Migrating
Mailbox From Exchange 2007 to  Exchange 2010: Very Slow                        
Exchange 2013 Mailbox Migration Rate/Speed ?                        
Public Folders after migration
to Exchange 2013 are extremly slow                        
Exchange 2010 mailbox move to Exchange 2013                        
Single Mailbox Very Slow after 2003
to 2010 Migration
I have also come across the most common solution which is written here
http://support.microsoft.com/kb/2807668, however this is not my problem.
My problem is that mailbox migrations are SLOW, I'm talking about < 15MB per minute.  I do NOT have any contentIndex errors on any of my databases, they are all healthy and stay healthy throughout the migrations.  Both Exchange 2010 and 2013
are DAGs.
When I 'view details' from the ECP migrations page, I see things like 'Stalled duration
00:53:21 ', however like I said there is no errors, and nothing mentioned in the user report.
Get-ClientAccessServer | Test-MRSHealth this checks out fine, no problems.
I have disabled offloading on all of the NICs.
Is there anything else I can do or monitor to find out why its so slow?
Could it be because of DAG even though there are no errors and the state is healthy.  Would an idea be to take down the dag during migration then add that after all is done?
Andrew Huddleston | Hillsong Church | Sydney