Discovering Under Utilized ESX Host in a production VMware environment using EM12c

Last week Blue Medora was pleased to release the Enterprise Manager Plugin for VMware. The plugin extends Oracle Enterprise Manager 12c (EM) to provide comprehensive monitoring and management of critical VMware-based virtualization infrastructures. Over the course of the next couple weeks, we will be blogging a number of user scenarios where the EM Plugin for VMware extends Oracle Enterprise Manager by providing deeper monitoring visibility as well as management capabilities. Below is a snippet of Blue Medora’s Enterprise Manager Plugin for VMware in action — a set of charts (pulled from the EM12c console) that display the distribution of ESX Hosts in a specific VMware cluster — broken down by CPU and Memory utilization.

Problem

The VMware admin has received requests from application owners to allocate additional memory to several virtual machines that were are members of Cluster-1. Unfortunately, Cluster-1 is already over-committed on memory.

Solution A – Buy more hardware

The VMware admin reaches out to their preferred hardware vendor to quote out additional hardware (either additional memory and/or more servers) to provide the additional capacity required to satisfy the application owner’s request for more memory allocation to their VMs.

Solution B – Use the EM Plug-in for VMware to evaluate capacity and adjust VM distribution

The VMware admin uses the metrics provided by the EM Plugin for VMware to help optimize the VM allocation within the VMware clusters. In the screenshot above we see the VMware plugin deployed on an EM12c OMA remotely monitoring another cluster, Cluster-2 (contains 10 hosts), in the same VMware vSphere v5.0 environment. Within a few minutes of deploying the plugin, the administrators quickly identifies under utilization of hypervisor resources in the monitored environment. Of the 10 ESXi hosts, 9 of them using between 0% and 25% of their available memory and CPU. There is obviously lots of excess capacity in Cluster-2. With this knowledge, the VMware admin took takes the following actions:

1) Two of the ESXi hosts in Cluster-2 were migrated to Cluster-1 to beef up capacity

2) Three of the ESXi hosts in Cluster-2 were also temporarily decommissioned — resulting in a short term energy savings

Results

• Cluster-1 is able to better serve critical applications
• Cluster-2 is now operating near capacity, but in a healthy range
• The administrator was able to postpone the next round of server purchases
• Energy costs decreased due to decommissioned hosts