When working with load balancers, we have a nice overview in the CloudWatch management console that already provides us with a nice set of metrics that we would typically be tracking with ELB. Possibly the most important part of monitoring ELB services is the ability to understand the load our application is under at any given time and the way it is responding to the load. A crucial metric to follow would be the number of requests and the latency of the response. We will be striving to maintain a low latency under a high load at all times, since the application user experience is directly tied to these two factors. When latencies rise upon an increase in the number of connections, our application is probably not scaled appropriately.

In the ELB metrics, we can also find the backend connection errors. We will either be looking at the combined number of backend errors, as we can see on the right-hand side of the following screenshot, or we can create a dashboard and collect 400 and 500 error metrics separately: