What you can't do is try to track servers' physical resources by monitoring the operating systems on the virtual machines. You must go below the hypervisor layer to track a physical server's resources separately from a VM's. You probably still need to track reported resource use by the VMs, but that's a separate issue and done for things like performance analysis on the apps the VMs are running. (Note: again, this assumes you're using hypervisor virtualisation like VMware or Xen.)
According to Anderson, memory bandwidth is usually the most critical resource on a physical server, simply because it is a hard limit. Unlike nearly everything else, from storage to processing cycles, you can't increase the memory bandwidth built into the box. When that saturates, you've got to move some of the VMs to a new physical server.
Bandwidth of all kinds is a particular problem because it is likely to be the place where problems show up. Not only do almost all applications use bandwidth, but the loads tend to be very variable. If two or more applications peak their I/O or memory demands at the same time, you're likely to have an intermittent problem that will be hard to track down.
This introduces, if not a new level of complexity, at least a new watch point in the virtualised data centre.
When the physical server supporting virtualised machines runs out of resources, a whole lot of applications are going to be affected. As Stucker puts it, "Anytime you miss your guess, there's going to be more delta in your peaks and valleys."
The need to assure the availability of resources is an additional reason why virtual/physical server stacks need to be monitored closely. Another reason is that a hardware failure will impact more than one application.
Because monitoring is so critical, virtualisation software like VMware and SWsoft's Virtuozzo include tools that help track both physical and virtual resources.
Because loads change over time, monitoring is a continuous process. You can't set it and forget it, any more than you can with a physical server. In fact, you have to pay even closer attention to monitoring, because each physical server is supporting several virtual servers and all of them can be affected if the physical server starts to run out of resources.
One peculiarity of virtualisation management is that a lot of companies aren't invested in third-party tools to do it. According to Bob Gill, managing director for servers at TheInfoPro, a New York-based market research firm, surveys show that most virtualisation users rely on the tools supplied by the virtualisation vendors rather than investing in additional applications to help manage their environments.
In part, this is because companies like VMware and SWsoft already provide management tools with their products. But part of it is also due to the relative immaturity of the virtualisation market. At the recent Infrastructure Management Conference, Gill said he expects the market for third-party monitoring and management tools to grow as the market for virtualised servers grows.