Hi, We have 3 Exchange Server 2010 boxes in a DAG, consisting of 7 DBs. Two of the Exchange Servers are sitting on our local site and 3rd is at our DR site. Replication takes place to the DR server and this works OK. We have about 450 mailboxes, averaging about 2GB each, max. The problems that I have having is with performance. Outlook 2010 clients are continuously showing the balloon popup "outlook is trying to send / receive data from your exchange server" and this can hang outlook for 5-10 seconds, maybe more. This can happen for up to 4-5 times a day for a user. Not all users are affected, but it affects all types of our users, such as Citrix XenApp users, PC users and laptop users. They all work in Exchange Online mode. The Exchange Servers are VMs hosted on Citrix XenServers (version 5.6 FP1) and the storage is a Compellent SAN. This is connected through iSCSI. There is a hardware load-balancer in place which is Citrix Netscaler, and this works perfectly fine. Have also changed the scenario and removed this and the problem is still present. I have done some performance monitoring on the Exchange Servers and the main point is the high Disk Read times on the DB mount points. These are coming in at an average of 30ms. I have checked on the SAN and it is not overly worked and has a cpacity of over 3000 IOPs. The average IOPs across all the LUNs on the SAN is 600 IOPs. The Exchange LUNs are averaging 200 IOPs. I compared this to another Exchange Server config that doesnt have replication and is a single Exchange Server, with 250 mailboxes. It also sits on a Compellent SAN. The disk read performance on this environment is alot less, and their users do not experience the same issues. The Exchange consultant that completed our installation thinks that the issue is with the SAN and the performance but I dont think it can be. There are 30 disks in the SAN, and all the LUNs are spread across the disks, so I would expect all the servers to have performance problems if the disks could not cope with the amount of IOPs being placed on them. Any help with this would be greatly appreciated. Thanks

i guess this issue due to user working on online mode... did you ever tried with cached mode for all the folks and see whether the same condition prevail.

I have tried changing some of the users into Cached Mode and it made no difference. That was one of the solutions I had thought of. However, it wouldnt a be a solution as 90% of our users are in a Citrix XenApp environment, so .ost files are not practical.

On Thu, 22 Dec 2011 12:44:16 +0000, AshPoxon wrote: > > >Hi, > >We have 3 Exchange Server 2010 boxes in a DAG, consisting of 7 DBs. Two of the Exchange Servers are sitting on our local site and 3rd is at our DR site. Replication takes place to the DR server and this works OK. We have about 450 mailboxes, averaging about 2GB each, max. > >The problems that I have having is with performance. Outlook 2010 clients are continuously showing the balloon popup "outlook is trying to send / receive data from your exchange server" and this can hang outlook for 5-10 seconds, maybe more. This can happen for up to 4-5 times a day for a user. Not all users are affected, but it affects all types of our users, such as Citrix XenApp users, PC users and laptop users. They all work in Exchange Online mode. > >The Exchange Servers are VMs hosted on Citrix XenServers (version 5.6 FP1) and the storage is a Compellent SAN. This is connected through iSCSI. > >There is a hardware load-balancer in place which is Citrix Netscaler, and this works perfectly fine. Have also changed the scenario and removed this and the problem is still present. > >I have done some performance monitoring on the Exchange Servers and the main point is the high Disk Read times on the DB mount points. These are coming in at an average of 30ms. I have checked on the SAN and it is not overly worked and has a cpacity of over 3000 IOPs. The average IOPs across all the LUNs on the SAN is 600 IOPs. The Exchange LUNs are averaging 200 IOPs. > >I compared this to another Exchange Server config that doesnt have replication and is a single Exchange Server, with 250 mailboxes. It also sits on a Compellent SAN. The disk read performance on this environment is alot less, and their users do not experience the same issues. > >The Exchange consultant that completed our installation thinks that the issue is with the SAN and the performance but I dont think it can be. There are 30 disks in the SAN, and all the LUNs are spread across the disks, so I would expect all the servers to have performance problems if the disks could not cope with the amount of IOPs being placed on them. > >Any help with this would be greatly appreciated. If you're averaging 30MS per I/O then you need to rethink your design. If the capacity of the storage is, as you say, 3000 IOPS (without benefit of caching in the SAN enclosures) then you probably have problem in your VMs (or the hosts). "Average" IOPS don't amount to much when you deal with perfromance problems -- what matters is the IOPS when you're having a problem. And, really, who cares about "all the LUNs"? All you should care about are the LUNs used by Exchange. --- Rich Matheisen MCSE+I, Exchange MVP --- Rich Matheisen MCSE+I, Exchange MVP

Thanks for the reply. I've tried to pinpoint the exact time that the Outlook hangs occurr but as they are never the same for all the users, i.e. not all users are affected at the same time. Also, some users just dont bother to report it to the Helpdesk any more.