I run vSphere in my personal lab, and I am constantly pushing my memory configuration to the max. To try to keep up with this moving target, I add more memory to the ESXi hosts every few months, yet I still find myself pushing the limits of my lab's capabilities.
One of the little irritations of this configuration is the host memory usage alarm; this kicks in when the host crosses 90% usage of memory to warning state and to alert state at 95% usage, sustained for five minutes in both cases. I would definitely keep this alarm active in production environments, but for lab or development environments, this value can be tweaked upward to reduce potentially unnecessary alarms for non-critical systems.Figure A shows the host memory alarm's default configuration. Figure A
Click the image to enlarge.If you enter a new number in the warning limit, it will identify when the alarm condition will engage again. In my lab environment, I am happy with using 98% for warning and 99% for alert levels. This new configuration is shown in Figure B. Figure B
Click the image to enlarge.
The magic here is that these values can be applied in a hierarchical fashion. The default values are stored in the parent container of the vCenter Server. Individual hosts or clusters can have their own values for alarms, so a development cluster can have a relaxed alarm threshold, while the production environment in the same vSphere data center can have the default arrangement.
Do you modify the alarm levels for some of the built-in vSphere alarms? If so, share your tweaks in the discussion.
Rick Vanover is a software strategy specialist for Veeam Software, based in Columbus, Ohio. Rick has years of IT experience and focuses on virtualization, Windows-based server administration, and system hardware.