One of the questions that has come to my mind is with all the additional features for power and performance management introduced with the Nehalem family of processors, how much will that affect the performance and efficiency of vSphere? I have been hearing a number of things from different vendors about their new server capabilities and I am curious about how important these new features are to users?
For instance, if you are using resource scheduling with vSphere and you can consolidate running workloads to fewer cores in off-peak times, how much more performance do you get with turbo-mode as you power down the surplus cores? And the corollary question, how much power are you saving directly and indirectly (from cooling, system resources, etc.) with the cores being quiesced?
The major vendors that I deal with point to the performance gains of 1 speed step for each core that gets powered down, so with 2 cores quiesced, the other 2 cores jump up 2 steps in performance which theoretically means that you can get a couple more VMs running on a fewer running cores under the turbo mode.
Also, in one example with the BL460c G6, HP is claiming that the combined power efficiency engineering they have been able to link to the new Nehalem based servers gives them a 20-40% power reduction over previous generations on top of the performance improvements. I have heard similar improvements from the other server companies, too.
Here is a Youtube video with a VMware engineer talking about DPM improvements in vSphere and how they are managing system resources better too: http://www.youtube.com/watch?v=LVNE5oqBMJ8
But do buyers and users really care about the features and functions beyond the improved performance and scalability? Please comment about how important these power management and performance management features really affect your interest and urgency in adopting Nehalem processor servers for WMware? And, what are YOUR expectations for gains from deploying these new capabilities?
