Monitoring Microsoft SCVMM

eG Enterprise provides a specialized Microsoft SCVMM monitoring model for SCVMM. Figure 5depicts this model.

Figure 5 : Layer model of the Microsoft SCVMM component

Each layer of Figure 5 is mapped to tests that employ PowerShell scripts/Perfmon to report useful metrics on SCVMM performance. With the help of these metrics, administrators can find answers to the following performance queries:

  • How many clusters are managed by SCVMM? Which ones are they? To which host group, does each cluster belong?
  • Is any cluster running out of CPU resources? If so, which one? Which host in the cluster is causing this CPU contention? Why is that host CPU-hungry - is because on/more VMs on that host are over-utilizing CPU resources? if so, which VMs are these? or is it because too much CPU resources have been set aside for OS use?
  • Is any cluster experiencing a memory contention? Which one is it? Which host in that cluster is low on memory? Is it because one/more VMs on that host are running memory-hungry processing? If so, which VMs are these?
  • Are all clusters using their storage resources optimally, or is there a cluster than is facing a storage space crunch now? Which cluster is it, and what type of storage is that cluster over-utilizing - remote? or local? Is any host in the cluster responsible for this space shortage? If so, which host is it?
  • Is Microsoft Network load balancer ntegrated with one/more service templates configured in the SCVMM?
  • Is network virtualization enabled for any logical network? If so, does the logical network use Generic Routing Encapsolution for network virtualization?
  • Are private VLANs enabled for any logical network?
  • Is any MAC address pool running out of addresses for allocation? If so, which pool is it and with which host group is that pool associated?
  • Which configuration providers are plugged into SCVMM?
  • Which port classifications are in use in SCVMM? What are their properties?
  • What are the refresher jobs that are running SVMM presently? Are any of these jobs executing slowly, causing the length of the job queue to increase? Or are many refresher jobs pending because hosts are responding poorly to them?
  • Is any host in a cluster not suitable for VM placement? If so, why? Is it because sufficient disk space, network, and I/O resources been reserved for the host?
  • Do any hosts need attention?
  • Is any host not responding to requests?
  • Are hosts in a cluster using their CPU resources optimally? Is any host CPU-hungry? Is it because, the VMs on that host are consuming CPU excessively? If so, which VMs are these?
  • Is live migration not supported by or not enabled on any host?
  • Are any VMs on a host powered-off? Which VMs are these?
  • Are there any orphaned VMs on any host?

To know which tests are mapped to the top 2 layers and what measures they report, use the links below:

The SCVMM Fabric Layer

The SCVMM Service Layer