Validating Load Balancers

Before you use test clients to validate the newly configured load balancers, it is a good practice to use telnet to confirm that the load balancer is at least listening on the virtual IP address (VIP) and directing connections to both Front End Servers. To do so, perform the following steps:

  1. On each Front End Server, run Netmon (or a similar protocol analysis tool) and add a filter to only display traffic that is coming in on TCP port 5061.

  2. From a test client, telnet to the VIP on port 5061.

  3. Close the telnet session and reestablish the telnet session several times. On each Front End Server, the Netmon trace should show TCP connections in an alternating fashion.

To validate the load balancer configuration by using test clients, we will employ similar configuration changes to the hosts file similar to those that were used to validate the Front End Server installation. The only difference is that the load balancer is now configured, and so we can point the pool FQDN to the load balancer VIP. This test simulates the behavior that will occur when the pool FQDN is updated in the production DNS server.

First, add a temporary hostfile entry on each newly installed Front End Server, pointing the FQDN of the existing pool to the IP address of the corresponding load balancer VIP. For instance, in our sample topology you would do the following:

  1. On ocsfe01.contoso.com, add a hosts file entry that resolves ocspool.contoso.com to 10.0.0.40.

  2. In a command window, run ipconfig /flushdnsto clear the DNS cache.

  3. Restart all services on the Front End Server to ensure that they use the updated IP address.

  4. Repeat the above steps on ocsfe02.contoso.com, setting it to the same VIP address.

Next, prepare three test workstations with the Office Communicator, Outlook, and Live Meeting clients, and then perform the following steps:

  1. Add a hosts file entry on each client that resolves ocspool.contoso.comto the corresponding load balancer VIP. In the sample topology, this would be 10.0.0.40.

  2. In a command window, run ipconfig /flushdnsto clear the DNS cache.

  3. In Office Communicator, configure connection settings to Manual, specifying ocspool.contoso.comas the internal server name.

  4. In the Live Meeting Client, configure connection settings to Manual, specifying ocspool.contoso.comas the internal server name.

  5. Using the test clients, sign in to Office Communicator by using some test accounts and verify that all peer-to-peer and conference modalities work, including IM, voice, video, and desktop sharing.

  6. Use one of the test clients to schedule a Live Meeting conference by using Outlook. Join the meeting and verify that all Live Meeting conference modalities work.

  7. Update the hosts file entry on each client to resolve ocspool.contoso.comto 10.0.0.42, and then rerun ipconfig /flushdnsto clear the DNS cache.

  8. Repeat steps 2 through 6 to validate functionality on the second Front End Server.

  9. If you discover any issues, troubleshoot and correct them before you proceed.

At this stage, you have validated that the functionality of the newly added load balancers works with the Front End Servers. By completing these validation tasks, you can be more confident when you perform the actual cutover to the production DNS. Before you do that, however, take some time to simulate failure of a Front End Server. For example, stop the Front End service, or pull the network cable. Identify the client behavior, and document the procedure for responding to such a situation. Clients typically use their inbuilt keep-alive mechanisms, client retry logic, and server reconnection randomizations to detect when their Front End Server goes down, and they will connect seamlessly to another Front End Server that is available behind the load balancer.

The final step is to remove the test settings that you configured:

  1. On ocsfe01.contoso.comand ocsfe02.contoso.com, remove the hosts file entry that resolved ocspool.contoso.com.

  2. In a command window, run ipconfig /flushdnsto clear the DNS cache.

  3. Restart all services on the Front End Server to ensure that they use the updated IP address.

  4. On the test clients, remove the hosts file entry that resolved ocspool.contoso.com.

  5. In a command window, run ipconfig /flushdnsto clear the DNS cache.

  6. Reset Office Communicator and Live Meeting clients to use automatic server lookup.

  7. Restart the test clients to ensure that they do not use any cached DNS entries.