Exchange 2013: Test system with Database Availability Group (DAG) and Kemp load balancers (Part 5)

In the last part of this series of articles, we test the configured HA environment. The environment has not yet been optimized, but the results are certainly impressive.

Here you can see that Outlook 2013 no longer connects to the Exchange Server FQDN, but to a value consisting of "Mailbox GUID + @ + UPN Suffix"

In the connection status we can see that all connections to mail.frankysweb.de are running

In normal operation (shortly before pulling the plug) everything still looks good, both Exchange servers are served by the load balancers

The mailbox database is online on server EX01

I simply unplug the power cables from the EX01 server. While I'm running back to see what's happening, the load balancers have already noticed that one Exchange server has failed and are only forwarding the requests to the remaining Exchange server.

You can also see in the ECP that the database has been switched. It is now active on EX02.

Outlook briefly loses the connection, but it takes less than 20 seconds until everything is back to normal.

The connection status changes back to "Established"

I run another test, restart the Exchange Server and switch the database back online to Server EX01.

After a short initialization, the database is online again.

The load balancers also forward requests to the EX01 again

I now switch off EX01 and the active load balancer, both at the same time. A continuous ping to mail.frankysweb.de loses 2 pings during the outage of the active page, then it continues.

The load balancers also notice the failure quickly. Even if the Exchange Server and the active load balancer fail at the same time, the connection is restored after just a few seconds. Outlook briefly loses the connection, but re-establishes it and work can continue.

With a few optimizations, there is certainly still room for improvement. But for a first test, it already looks very good. Mission complete.