Performance forms the basis of measurement of any system, process, or solution that is performing an action. How good the performance of the system is tells us how effective and beneficial it is in meeting its intended purpose. Monitoring performance by measuring performance indicators is the right way to keep performance under control. Performance monitoring is not the same for all aspects of the IT infrastructure. A virtualization environment (VMware or Hyper-V, etc.) differs from a server or application environment in the context of performance management. There is more to monitor with virtualization: different entities, different metrics interfacing between the physical hardware and the guest operating system. What makes virtualization performance management difficult is that the guest OS is not seeing the physical hardware, and instead seeing the virtual hardware that is emulated by the hypervisor. To gain complete visibility, what you need is a VMware and Hyper-V monitoring solution that is aware of the virtualization layer and can distinguish it from the physical hardware and exactly pinpoint the source of an issue.
The performance of a virtualization environment is best measured by capturing a combination of VM and host statistics such as:
- CPU usage and the time taken for a CPU to be available
- Memory actively used, allocated and swapped by the VMkernel
- Disk usage in terms of I/O, latency to process command issued by the guest OS
- Network in terms of the number of transmit and receive packets dropped
In a virtualized environment, pinpointing the source of a problem can get complex as the symptoms can be caused by many different things. The very nature of a virtual environment – which is all about the sharing of resources by hosts and VMs and the flexibility to move VMs – can make troubleshooting performance problems more difficult. For example, a host performance issue can cause the VM resource contention and performance bottlenecks. A VM performance issue may lead to slowdown of applications running on it.
To stay ahead of performance problems, it is essential to detect the underlying resource contention and the top resource consumers in the environment, and also to be able to spin up performance trends to get a forecast of where the virtual environment is headed in terms of resource utilization, capacity and workload assignment.
End-to-end and holistic virtualization monitoring – from host to VM to storage – is the best solution that helps get insight into performance metrics of your virtual infrastructure. If you need to get ahead of virtualization performance problems, you need to be able to receive real-time alerts pointing out irregularities, so you can take corrective actions immediately. In its essence, performance management is a must-have capability to constantly monitor the health of the virtualization infrastructure, identify issues in real time and prevent them causing virtualization performance bottlenecks and eventually application performance issues.
Read this free white paper to learn more about virtualization performance management.