A super metric is an administrator-created custom metric that contains a combination of one or more metrics for one or more objects. A super metric can be derived from either a single object or multiple objects across multiple environments.
To understand the use of super metrics and its functionality let’s take an example.
An administrator notices high CPU ready value on a virtual machine, and wants to know if it is the case across all VMs on the host or vSphere cluster. The administrator might check what the average CPU ready value is for all VMs across the cluster.
Unfortunately, a suitable metric to perform this analysis is not available on a vSphere host or cluster (other useful metrics may be available as workaround), this forces administrator to look through multiple VMs individually to get this data.
Here administrator could create a super metric that rolls up the average CPU ready value of all VMs to the cluster level, saving a lot of time by not having to view each object individually.
Super Metrics Terminologies:
Objects and Object Types:
An object is a single entity in vROps such as a virtual machine, a datastore, a switch, and so on. In previous versions of Operations Manager, an object was referred to as a resource.
An object type is a group of objects that share the same set of properties and metrics. This can also be seen as a category and includes examples such as virtual machine, datastore, and vSphere Distributed Switch.
Object types are the most common selection, as they allow super metrics to be applied to all objects of the same object type with a single formula, rather than creating individual formulas for individual objects. Object types were previously referred to as resource kinds.
An object has metrics that are added to the FSDB and can be retrieved on request. Object includes examples such as VM CPU co-stop (ms), VM Host Disk Usage rate (Kbps), and VDI desktop pool average PCoIP latency (ms).
Metrics are used in all areas of vROps including dashboards, views, widgets, and alarms. An individual metric always has a 1:1 mapping with its parent object.
Attribute types (previously known as attribute kinds) are the definition of the array of metrics an object type. Attribute types are commonly used in super metrics as they allow the formula to be applied to more than one object.
Following are the three main types of super metrics:
• Generic resource
• Specific resource/Pass-through
A rollup super metric uses functions such as sum, avg, min, count, and so on, and then applies them through a looping function of all child objects matching the object and attribute type. The result of this calculation is then available as a super metric on the object where the super metric is applied.
This figure shows an example of two different super metrics of the rollup type applied at different levels. There are two different ESX hosts (A and B). On host A we have 1200 GB overall free space and on host B has 600 GB free space. The administrator wishes to know at a cluster level how much free space is remaining. The administrator creates a super metric that uses the sum function to loop through all instances of free space for the datastore object type.
This super metric is then applied to the vSphere cluster via a policy, and the result of 1.8 TB is given by the super metric.
A generic resource super metric is used when you combine existing metrics using mathematical operators and apply them back to the same object or another object with the same attributes. This essentially allows an administrator to create a completely new metric for an object type where a gap has been identified. What makes a generic resource super metric different from others is the use of the This option.
A specific resource or pass-through is a super metric that is targeting a particular object rather than an object type. It is also referred to as pass-through, as a particular object is being targeted and this super metric can be applied anywhere without the need to specify depth as with the rollup super metric.