Client-Side Monitoring with Targeted Groups and Load Balancers
Updated: May 13, 2016
Applies To: System Center 2012 R2 Operations Manager, System Center 2012 - Operations Manager, System Center 2012 SP1 - Operations Manager
When load balancers are used during client-side monitoring, the load balancer is designed to get the true client IP addresses. In a simple case, when a client is connected directly to one web server, the web server knows the client IP address. However, when you have several servers and use a load balancer to distribute traffic from the clients among the servers, this can present two problems. First, the IP reported to the collector is the virtual IP that the load balancer uses, not the real end-user IP. When users hit the load balancer, it sends them to an appropriate web server. Due to the load balancer, the web server sees only the internal port IP (inside the network), not the real IP (outside the network). Additionally, if some servers are monitored and some are not, data can be lost because responses from the clients are often sent to servers in a least load or round robin fashion, which means that the response might go to a server that is not hosting a collector.
Here are some strategies for setting up client-side monitoring when you have multiple monitored servers and use load balancers.
Solution 1 If you are authoring a new .NET Application Performance Monitoring template and including client-side monitoring for a targeted group, we recommend that you choose a group of servers that are all served by the same load balancer. Target monitoring to all servers in the load-balanced farm.