I work with cloud computing and elasticity, and I need to study metrics to measure elasticity and realize performance analysis. Specific or indirect metrics.
Everything depends on the Application which you have chosen.
Some Application utilizes Memory or CPU or disk or Network more or may consume all.
The above metrics [CPU, MEMORY etc]. shall be considered as metrics.
Also you have to chose what type of elasticity you wish to chose [Horizontal or Vertical].
If its horizontal, latency between Front end and back end need to be considered.
If its vertical scaling, Physical Server's Capacity is required to be considered.
From Elasticity, You shall able to determine time at which heavy traffic was present and time at which less traffic is available;which helps in better capacity planning.
If you target web applications or continuously running apps that consume and produce messages you may want to look at the latency of the response. An application that does not scale will exhibit long delays in getting a response if numerous requests are hitting it all of the sudden. An elastic application will scale out and will reduce the latency. By measuring latency at app level you will also have the advantage of including the VM boot times in it.