Virtualization performance monitoring is critical for diagnosing problems and planning for future capacity. VMware vCenter Operations, announced in March, is a virtualization management tool in the vCenter suite that provides performance monitoring for your VMware infrastructure.
Normally, a virtualization administrator would have to understand and correlate performance monitoring statistics against one another and over time. VMware vCenter Operations, however, uses a complex algorithm that analyzes statistical points across the entire vSphere infrastructure. It distills this virtualization performance data into three easy-to-understand "badges" for each object in the virtual machine (VM) tree.
Each of the badges in VMware vCenter Operations displays a number and a color indicator for virtualization performance stats. The color changes from green to yellow to red as an object's resources approach unwanted or dangerous values.
Workload. This performance monitoring badge is a measure of the utilization of resources by a given object. In VMware vCenter Operations, it is displayed as a percentage of the resources the object needs versus the amount of resources to which it has access. A value of 100% indicates that all the resources the object requires are satisfied, but it has no extra resources available. Values less than 100% are better because they indicate that the object has more resources available to it than it needs. The object is being starved for resources if the value is greater than 100%.
For example, a VM that has all its resource needs satisfied has a value of 100% or less, but if resource overcommitment occurs, the needs of this VM can no longer be met and the value will be greater than 100%.
Capacity. This badge indicates how close an object is to resource exhaustion. Based on historical performance monitoring trends , VMware vCenter Operations calculates the amount of time remaining before the current capacity of the object is no longer sufficient.
For instance, say VMware vCenter Operations is monitoring a cluster with overcommitted RAM. The VMs in this cluster are receiving all the RAM they require and virtualization performance is good. Based on past utilization trends for several VMs that consistently increase their RAM utilization, VMware vCenter Operations can calculate when the VMs will require more RAM. When they do require more, the value of the Capacity badge will decrease for the cluster.
Health. This badge displays the relative normality of the current state of an object. This performance monitoring statistic is based on historical trends for the object. A value of 100 indicates normal operations. As the object’s behavior deviates from what’s expected, the value will decrease.
For example, a database server for the IRS may not be heavily utilized for most of the year, but in April the server’s utilization increases drastically. This spike may trigger alarms due to high utilization or insufficient resources, but the Health badge will know that historically this is normal for this period of time. The Workload badge may go into a red status, but the Health badge will show nothing unusual.
VMware vCenter Operations reduces a very complex set of performance monitoring counters down to a simple set of indicators. The layout of the virtualization performance information also makes it easy to see how parent, children and sibling performance may be affecting a particular object.