Feeling a VM Resource Squeeze? Practice Active Capacity Management

Cloud infrastructure is all about providing the right amount of resources for your applications at any given moment. Overprovisioning might be wise for performance-oriented apps, but generally “right-sizing” is the best way to maximize your budget, especially as most IT departments face efficiency and cost struggles.

By being proactive about managing your virtual machine resources and halting underutilized or “zombie” VMs, you can free up those resources either to be decommissioned or reassigned to other uses.

You’ll want to adjust VM size to reclaim overprovisioned VMs, clean up idle or turned-off VMs, and resize VMs that are stretching their current resources beyond acceptable performance. Here’s how to practice active capacity management.

When to practice proactive capacity management

Only production VMs that are well into their lifecycle should fall under an active capacity management plan. These are virtual machines that are running applications on a constant basis. They’re relatively stable and haven’t been tweaked in size or configuration for some time.

If your virtual data center is experiencing serious VM sprawl or you have noticeable performance issues, now is the time for capacity management. If you’re facing a budget crunch or need to free up some virtual resources for an upcoming project, you’ll want to undergo an active capacity management overhaul. Once you establish a workflow, proactive capacity management on a regular schedule will keep your cloud infrastructure in top shape while minimizing expenses.

How to proactively manage cloud capacity

The first step, as with most any cloud project, involves pulling reports to see the scope and performance of your entire virtual data center. You can do this individually under the Performance tab in vCenter, a third party reporting and monitoring product, using PowerCLI scripts, or with vRealize Operations Tool. This should be performed by someone with extensive understanding and experience with your cloud.

Analyze this data by sorting it by name, data center, what piece of your business or organization it is related to, and how it is performing. Note if the CPU or memory for each are oversized or overstressed. Record all turned off or idling VMs. From here, you can create a ballpark of operational expense adjustments – how much can be saved from overprovisioned VMs and how much you’ll need to spend on stressed VMs.

Create a plan to adjust VMs on a fixed schedule, including backups/snapshots and freezing. Make sure the departments who depend on each application know when their applications might be affected. After you edit each VM, use the tagging feature of vSphere to note that they have been adjusted (with a date). Monitor adjusted VMs for several weeks, also on a fixed schedule, to ensure they are performing as expected. When you’re sure that each application is working appropriately, report on your savings or other adjustments.

Communication is key throughout this entire process. Keep all stakeholders and management involved and informed about the process as you adjust the resource capacity of your virtual machines. This not only lets people know when services might be spotty, but also justifies your work and department as a whole. Be sure to project your savings out over the long-term, as reclaiming resources to right-size your data center adds up significantly over time, while also improving performance of production VMs.