With many organizations choosing virtualization for their IT systems, there are more considerations for data center managers to find a balance between energy efficiency and performance. A single server may host a number of virtualization clients with loads increasing and decreasing sporadically.
According to an article in CIO Asia, metrics such as Power Usage Effectiveness (PUE) can be misleading in virtualized environments, and efforts to improve PUE at the server level can become excessively complex and counter-productive. It’s better to combine device-level optimization with a holistic system-level view, perhaps using a Data Center Infrastructure Management (DCIM) tool.
The CIO Asia article advises the data center manager to use DCIM in four steps:
- Audit the deployed equipment and its performance. Preferably, the DCIM will have an intuitive interface for capturing all data points associated with each asset such as size location, power consumption and cooling needs.
- The next step is to develop strategies. Typically, the group of low-performing servers are decommissioned, and the group of highest-performing servers aren’t a problem. For the mid-to-low range servers, the manager must decide which servers to virtualize while trying to balance efficiency motives and honoring performance requirements. The challenge is to define reasonable power capping policies for each category of servers, according to the article.
- Step three involves testing possible strategies. Start with the least aggressive policy and verify its results before moving on to more aggressive policies, advises the article.
- Finally, after all policies have been applied, the data center manager can tune the physical infrastructure, using the DCIM tools.
Photo: Servers via Shutterstock