One of the main advantages with introducing virtualization into an environment is the ability to logically provision compute resources on demand. This allows for a tremendous amount of flexibility in dynamically scaling up to meet the needs of a workload and to ensure maximum utilization of physical servers, but what happens when resource utilization goes down? Does it make sense to leave your entire environment powered on thus consuming expensive electrical and cooling resources when the ESX nodes are essentially sitting idle?
VMware vSphere, or ESX 4.0, introduced the ability to dynamically migrate virtual machines to a subset of hosts as utilization drops and then power off nodes to conserve on power and cooling. This feature, called Distributed Power Management, relies upon the Intelligent Platform Management Interface, IPMI, to integrate with the physical server layer of the environment and make this functionality possible.
Cisco UCS has taken their purpose built, virtualization platform to a new level by enabling integration with IPMI and therefore allowing for use of DPM when running ESX on UCS blades. Together, these technologies can take efficiency to another level, especially in environments where utilization may scale up or scale down drastically and on a regular basis such as is the case with VDI and various other ASP models.
The following whitepaper outlines the steps necessary to enable DPM in a Cisco UCS environment running vSphere. Note, the Distributed Power Management feature requires vSphere Enterprise or Enterprise Plus licensing to enable this feature.
http://communities.vmware.com/servlet/JiveServlet/downloadBody/11874-102-1-1172/Integrating%20Cisco%20UCS%20with%20vSphere%20Distributed%20Power%20Management.p
0 comments:
Post a Comment