Use a combination of DRS rules and HA tags to manage workloads

Together, Distributed Resource Scheduler and vSphere High Availability play a major role in any modern VMware virtual environment, and as more critical workloads come online, it's essential that administrators carefully manage and document these important features. In order to use Distributed Resource Scheduler for load balancing VMs and workloads in VMware clusters, you must first establish a set of rules that dictate where it places VMs on a host.

DRS rules to keep or separate VMs have been an important part of DRS for some time now, but VMware recently added the ability to balance the quantity of VMs across hosts and groups. If someone chooses to deselect the DRS option rather than disable it for maintenance, it can easily wipe out these DRS rules. Although this won't cause any immediate outage issues, it makes the administrator's job harder because DRS rules are often complex and difficult to replicate, especially if you don't have thorough documentation. This happens more often than you might realize, even to veteran administrators.

This is where VMware tags come into play. Tags are nonoperational labels you can assign to VMs, hosts, networks, resource pools and more, and they can be a true lifesaver. You can use tags to create categories, such as whether to keep VMs together or separate. As you create tags, you have the option to change the tag name and description field of the VM with which it shares a rule or rules. Perhaps the best thing about tags is that they appear on a VM's summary page. It's possible to document DRS and vSphere High Availability (HA) rules in a spreadsheet or with Visio, but the simple truth is that this information will eventually become outdated. If an admin has to go to several different locations to create documentation, the risk of error and forgetfulness only increases. In addition to being easily accessible within the VMware client, tags are also part of the VM's data profile. These tags become searchable fields and are even accessible through PowerCLI commands and other tools that can see into the vCenter database.

The same applies to HA rules and what you decide to do with both isolation and restart priorities. HA and tags allow you to document which VMs are powered on or off and which VMs take priority. This takes the guesswork out of the recovery process after an outage. HA settings can be just as complex as DRS rules and just as fragile: If someone accidentally deselects an HA setting, it will uninstall rather than turn off the host monitoring aspect. You can create tag categories for the HA level of your VMs and simply assign these categories to each VM as it's created.

The true power of tags isn't what they do but where they're located. With everything that happens day to day in a data center, putting critical information in different applications can cause serious confusion in the event of a disaster. Virtual environments are complex, and with so many possible settings, it isn't a matter of if your staff will become overwhelmed -- it's a matter of when. Knowing which settings apply to each VM can make a difference not only in management, but also in recovery and compliance. Each VM comes with a description field, but you can do more with tags, from filtering to using PowerCLI. Keep in mind that tags are only helpful if they're used consistently. It's critical that tags become a fundamental part of the VM creation process because, if you miss even a few tags, you jeopardize the consistency of the environment.

As a final note, the function of your tags might vary depending on which client you use. At present, the Flash-based client has more functionality than the HTML5 client, but that's likely to change over time. So, don't let that discourage you from getting started with tags.