Won't using Distributed Power Management to power down a server cause management software to report an error? If so, is there a workaround?
Server management software can indeed report false alarms when Distributed Power Management (DPM) shuts down a server. It's an unintended consequence of some data center monitoring tools like BMC Performance Manager, CA Virtual Performance Management, Microsoft System Center Operations Manager and others. This issue is often overlooked until DPM starts to shut down servers.
The problem occurs because the management software tools are not aware or orchestrating the server's state, so the tool cannot tell the difference between a planned shutdown -- using DPM -- and some other unplanned failure such as a damaged server power supply. As a result, the management tool will typically send major alarms to administrators and perhaps even attempt to invoke other workload balancing actions in response to lost communication with one or more hardware platforms.
Generally, you can take two actions to help integrate DPM with other systems management tools. First, disable any management software alarms generated when a system is powered off using an Advanced Configuration and Power Interface (ACPI) S5 state. If some level of insight is desired, reconfigure the alarm to a low severity where no action is required. Management tools can usually monitor this through the vSphere API. Second, it is highly recommended to flag an alarm when DPM is unable to power-on a server. If the server cannot return from power-off, a serious problem may exist that requires investigation and correction. This can also be monitored with the DrsExitStandbyModeFailedEvent through the vSphere API.
Virtualization has matured to the point where organizations can look past the obvious benefits of static consolidation and take advantage of emerging tools like DRS that can intelligently automate consolidation and computing resource allocation based on workload activity levels. When properly coupled with power management utilities like DPM, virtualization can optimize workload placement and actually shut down idle servers to save the maximum amount of energy -- further lowering the cost of enterprise computing.
Dig deeper on DRS and DPM
Related Q&A from Stephen J. Bigelow
We use an IBM System z mainframe and it's time to update some aging tape drives. How big of a difference is there between the tape storage products ...continue reading
Tape, disk, virtual disk and other storage media all work with my mainframe, but how do I decide which is the best option for storing data?continue reading
I want to share a tape drive between the IBM System z mainframe and other IBM systems. Are there prerequisites for tape drive sharing?continue reading
Have a question for an expert?
Please add a title for your question
Get answers from a TechTarget expert on whatever's puzzling you.