Blade Server Power Study - IBM BladeCenter and HP BladeSystem
This report presents the results of tests conducted by Edison Group to compare the power consumption of the IBM BladeCenter blade server system with a comparable BladeSystem blade server configuration from HP. It provides background, configuration details, and methodology by which Edison reached its conclusion: that IBM BladeCenter H requires nearly 10 percent less power than the equivalently configured HP BladeSystem c7000.
Objective
Edison sought to address the lack of any industry study reflecting high-volume blade server configurations representative of those commonly deployed today within customer installations. We developed tests to establish parity between IBM BladeCenter and HP BladeSystem setups in a lab setting and measure the actual power consumption of each.
Many of the components within various commercially available servers —processors, memory, fans, power supplies, etc. — are supplied by the same manufacturers. For this reason, no appreciable difference in power consumption exists between servers at the component level. However, IBM BladeCenter as a whole (blades and chassis combined) is purposefully designed to maximize the efficiency of power delivery and cooling. Edison sought to accurately measure the power consumption of the IBM BladeCenter platform as compared to that of HP’s BladeSystem. Therefore, we benchmarked a workload on complete assembled systems.
NOTE: IBM refers to the structure in a blade server that contains the blades as a “chassis,” while HP refers to the same thing as an “enclosure.” For the sake of clarity we use the term “chassis” throughout this report.
Get pdf Blade Server Power Study - IBM BladeCenter and HP BladeSystem
