Monitoring VMware vCenter
eG Enterprise prescribes a specialized VMware vCenter monitoring model (see Figure 1), which can be managed in an agent-based or an agentless manner, and can be configured to periodically check the health of the critical services offered by vCenter, so as to proactively alert administrators to real/potential anomalies.
Every layer depicted by Figure 2 is mapped to a wide variety of tests that report the following:
- Is vCenter available? If so, how quickly is it responding to requests?
- Does vCenter possess adequate ESX server licenses?
- Is vCenter overloaded with sessions?
- Are sessions logging out smoothly, or are there any unexpected log outs? Which are the user sessions that logged out suddenly?
- How may sessions does a user have open on VC? Which user has the maximum sessions open on VC? Can I know the details of these sessions?
- Have any VC tasks failed? Which ones are they?
- Which are the tasks that have succeeded?
- Is any VC task taking too long to complete?
- Are there any pending tasks on VC?
- How many ESX servers is VC currently managing? How many of these were added recently?
- How many ESX servers were removed from VC?
- Are any ESX servers disconnected from VC? Were there any recent disconnects?
- Are any ESX servers not responding to VC? Did any ESX server become unresponsive recently?
- How many clusters have been configured on VC? How many ESX servers are managed per cluster?
- How many VMs are available in each cluster, and what is their current state? Are there any powered-off/suspended VMs in a cluster? If so, which ones are they?
- How are the clusters using the resources allocated to them?
- Is any cluster consuming resources excessively? Which resource pool on the cluster is responsible for this? How many VMs and child resource pools does this resource pool consist of?
- What are the datacenters managed by VC? What is the current configuration of each datacenter – i.e., how many ESX servers exist with a datacenter? How many VMs, clusters, networks, and datastores are available for the datacenter?
- How many ESX servers and VMs are directly assigned to a datacenter, and how many operate within a cluster?
- Which datastores have been configured for use of the virtual infrastructure? Which ones are currently available?
- How many ESX servers and VMs are using a particular datastore? Which ones are they?
- Is adequate free space available with the datastore?
The answers to the above questions can shed light on many key performance aspects of vCenter, and enable administrators to accurately pin-point the root-cause of problems with vCenter.
The bottom 7 layers of the monitoring model are similar to that of the Windows Generic Server model. You can refer to the Monitoring Unix and Windows Servers topic for the details on the tests mapped to these layers.