I have a three member Exchange 2010 DAG.  Two members (MB1,MB2) are in Site A, and the third member (MB3) is in Site B.

I have a Riverbed Steelhead WAN Optimizer in both Site A/B.

I'm replicating from Site A to Site B.

I can see that replication traffic is going through port 64327 on MB1 and that ports 59090,59091,59092 and 59111 used on MB3.  I have a connection for each mailbox database.

Compression is disabled in the DAG.  We have a 100MB MPLS.  We are getting 300MB to 400MB LAN throughput with the Riverbed using the full 100MB MPLS.

Now when I add a 5th mailbox database copy from MB2 to MB3, I see that port 64327 is being used on MB1, but the port on MB3 is 16125.  Now, I only see 6MB being used in the MPLS instead of closer to 100MB.  Riverbed is still providing the same reduction rate.

I saw the port change on MB3 from 59000 to 16000.  Is this the issue?

Anything else to look at?

Thanks

Ron

You might want to ask in a Riverbed forum.  Honestly, I've never had a customer that deployed a Riverbed where it didn't screw

Will do.

Thanks

Hi TrojansBaby,

What you need to ensure is there is no restriction, filtering, scanning, port restricition between the two Exchange servers.

Latency should be within the desired limit of 250ms.

DAG across the WAN:

https://social.technet.microsoft.com/Forums/exchange/en-US/99a3fef5-5798-4a40-a8ff-85e5252cb10d/dag-across-the-wan

Do you see any errors, there need not be an issue, if WAN bandwidth is not utilized.

Hi TrojansBaby,

Thank you for your question.

The port of DAG replicate network  on hosting active database cope mailbox is 64327, and a random ephemeral port was used on hosting passive server. It is by design.

The more detail could be referred by the following link:

http://blogs.technet.com/b/timmcmic/archive/2014/05/06/exchange-2010-2013-what-constitutes-a-failure-of-the-replication-network.aspx

I thing MPLS will limit the bandwidth, it will save the bandwidth for other services, the more details you contact Riverbed.

If there are any questions regarding this issue, please be free to let me know.

Best Regard,

Jim

Thanks

Riverbed is anaylizing system dumps to determine what is happening.

Ron

After additional trial and error, the underlying issue was networking.  A static route was missing.

Thanks

Ron

Hi Ron,

Thanks for coming back, you can add some more details on the solution and mark it as an answer.

After additional trial and error, the underlying issue was networking.  A static route was missing.

In our environment, we have two MPLS networks for all of our offices except our colo.  We just recently relocated to our new colo where we only have one MPLS provider.

The DAG network on the source side didn't have a static route to point the DAG network on the source side to the DAG network at the colo.  Once the static route was added, DAG replication started working consistently.

That's it.

Ron

• Marked as answer by TrojansBaby 13 hours 27 minutes ago

After additional trial and error, the underlying issue was networking.  A static route was missing.

In our environment, we have two MPLS networks for all of our offices except our colo.  We just recently relocated to our new colo where we only have one MPLS provider.

The DAG network on the source side didn't have a static route to point the DAG network on the source side to the DAG network at the colo.  Once the static route was added, DAG replication started working consistently.

That's it.

Ron

• Marked as answer by TrojansBaby Friday, September 04, 2015 6:04 PM

Thanks Ron! Appreciate it.