Sashkin - Fotolia

Do we need a unique VM lifecycle management plan for every workload?

When managing virtual machines, each one should follow the same general settings but it's important to make sure the particulars are tailored to each individual workload.

Proper lifecycle management is a critical goal of any virtualized infrastructure, allowing an enterprise to understand which workloads are deployed, the resources being used, the performance of each workload, the business value of each workload and the anticipated service life of each VM. When looking at VMware lifecycle management, one big question is how to treat individual virtual machines. Some might think each VM deserves its own settings while others think one set of rules fits all.

Is it really important to subject every VM to the same lifecycle management strategy? Can we treat VMs differently?

Although it may seem like these questions should have the same answer, they don't. Yes, every virtual machine should obey the same overarching lifecycle including justification, provisioning and protection, monitoring and optimizing, and review and decommissioning. This overall process ensures only VMs that impart value to the business are created with business resources, that the VMs are protected and managed properly and the VM resources will be available for re-use when the VM is no longer needed.

However, the attributes applied to various phases of the overall lifecycle can vary considerably depending on the workload's importance. For example, administrators may choose to overprovision a VM intended for a mission-critical workload (such as a database) to help guard against potential performance glitches during periods of high demand, and management alerts may be configured to warn administrators long before resource use becomes critical. By comparison, less-important VMs may receive fewer "extra" resources (tighter provisioning) and VM lifecycle management alerts may be set much closer to resource limits -- there is less to lose if the VM's performance falters a bit.

Similarly, the data protection scheme applied to each VM can also vary in proportion to the particular VM's value. For example, VMs running mission-critical email server components might receive copious failover planning, frequent snapshots, replication to several off-site locations and even be integrated into a high-availability VM cluster. By comparison, less-critical workloads might receive only infrequent snapshots and little, if any, failover preparation because it's not a big deal if that workload becomes unavailable for a time.

So while the same VM lifecycle management approach should ideally apply to every VM, the management, monitoring and other particulars can be tailored to a specific workload.

Dig Deeper on Virtual machine performance management