At times, VMware administrators go through the motions of daily VMware maintenance tasks. Admins do X, check Y...
and keep a finger on the pulse of Z. But it's important to reflect on why we carry out these maintenance tasks.
But maintenance isn't just about sustaining uptime. It ensures that you get the most out of your investments. ESX, vSphere, vCenter, and Workstation are well-established virtualization technologies, but they still require routine maintenance.
Take the time to understand your infrastructure, and create a VMware maintenance schedule to meet service-level agreements and minimize job stress. The following guidelines cover some best practices, but every virtual infrastructure is different, so determine what works best for you.
VMware maintenance tasks can be roughly broken down by frequency: daily, weekly and monthly.
Daily VMware maintenance tasks
I carry out the following maintenance tasks on a daily basis:
- Monitor mailbox for alarms. Mailbox monitoring is ongoing, passive and, once the appropriate alarms are configured, doesn't require much effort. As you become more familiar with your infrastructure, you can differentiate between alarms that require immediate attention and ones that are indications of a gradual baseline shift.
If a particular database job triggers CPU usage alarms at the same time each day, it's wise to adjust your alarm tolerance and frequency. Remember "The Boy Who Cried Wolf" Aesop's fable? Don't let persistent alarms lure you into ignoring important ones.
- Visit the server room. I have alerts configured, but I can tell exactly what is happening after a quick visual of a host's component displays -- which is helpful, especially if I miss an alarm or alert.
Are all the fans running? Are the memory sticks error-free? Do the storage area network (SAN) drives look good? I look at the hardware light-emitting diodes for a basic status update. I also quickly check the uninterruptible power supplies' voltage, as well as the current runtime. From that information, In case of a disaster, I know how much time I need to power off equipment. In addition to these quick visuals, I often use Hewlett-Packard Co.'s integrated Lights-Out (or iLO) ports on ESX hosts to check the hardware status, including temperature levels -- which is mandatory if you don't have physical access to the hosts or SAN.
- Look around vCenter/VirtualCenter. I check out any tasks that haven't been completed, glance at the ESX host performance and get a feel for how everything is working. When you are familiar with how the systems perform on an average day, it's easier to isolate problems. And believe me, problems will emerge at some point -- no matter how well your system is tuned. This process is similar to knowing your resting heart rate, and periodically checking it on a treadmill or bike.
Weekly VMware maintenance tasks
Each week, I perform the following activity:
- Back up vCenter/VirtualCenter database. My infrastructure does not change often, so I perform weekly database dumps and full backups of my management server. If your infrastructure is more dynamic, you should do these tasks more frequently. It cannot be emphasized enough: In case you need to rebuild, have good database backups.
Monthly VMware maintenance tasks
I carry out the following maintenance chores each month:
- Clean up storage. If there are any unnecessary snapshots, it's a good idea to get rid of them. If you are unsure whether you have old snapshots, you can find them with VMware's SiteSurvey.
- Revisit support agreements. Are your support agreements up to date? Is it time to start writing purchase orders to make sure you have proper help?
- Envision future improvements. So far, the focus has been on keeping the environment functioning. Take a step back, close your eyes and just think about what you would like the environment to do, instead of the other way around. How can it make your business better? Map out a way to get there. It sounds like daydreaming, but it may be some of your most productive time all day.
About the expert