Average Metric From A Group Of Virtual Machines

Is there a way to display a graph and also raw data for the average metrics across a group of VMs?  So say I have a group of web servers dedicated to an application and I want to know as a whole what their average CPU, Memory, etc is. 

I thought great, I just create a service and this should be a snap.  Well I still struggle as to why there is this great way to create services but the information that can be derived from them is so limited.  If I created a service for each layer of VMs in an application, it should really easy to report on metrics on the service just like an individual VM or even for each of the individual VMs associated with the service for that matter.  If this isn't what services can do, can someone explain what benefits there are to a service?  So far all I can get out of them are SLAs and with a recent cartridge, the ability to report on recently created VMs for different business units (thanks for that cart btw!)