Strategies for Using Intelligent Power Management to Reduce: TRANSCRIPT

Rob Enderle: Good to be here.

Karen Guglielmo: Great! So, to get started, I know there has been a lot of talk about the green data center over the last few years. In your opinion, is it still a big problem?

Rob Enderle: Well, it's a huge problem. I mean, what you have is a situation where folks have reached a capacity in terms of their data planning. They've got a data center that can handle only so much power, only so much heat, yet they've got to add more capacity as their company grows. The end result is this, they have to restructure the building, add more cooling, add more equipment, a very costly process that goes on top of buying the server and makes things incredibly expensive, not to mention the fact that every company has a green initiative, they are trying to be more cognizant of the environment and keep their carbon footprint lower, they want to do the right thing.

Karen Guglielmo: Okay, and from your experience with talking with customers, what are they telling you are their biggest pain points relating to power and cooling issues in the data center?

Rob Enderle: Well, it's a capacity issue. The data center is only so large. Its already probably been at capacity for some time, and to add that extra performance what they've got to do is either expand the data center, or more important, expand the cooling that has been delivered to the data center through air conditioning in Keller all of which costs significantly more than the server capacity we are thinking of adding. So, they are between a rock and a hard place. This is a tough environment in terms of the economy. They don't have a lot of money and they are trying to do things the most efficient way, the most inexpensive way possible, yet the way there were structured just doesn't allow them to do that with old technology.

Karen Guglielmo: So, is reducing power consumption in the data center a priority today for the customers that you are talking to?

Rob Enderle: It's a huge priority. The...their at capacity...the problem that they have got is one that is more physical than technological, they simply have got not enough capacity, not enough thermal capacity in the data center to handle the loads they need for the extra performance they require to grow their overall business and so they are between a rock and a hard place. They need a solution that will, you know, give them the performance they need, but still fall within the thermal capacities they already have.

Karen Guglielmo: Okay. So, I understand that Intel has an intelligent power management solution to address some of these problems we've discussed. Can you tell me what it is and how exactly it works?

Rob Enderle: Well, Intel Intelligent Power Management kind of cute phrase, the...what it does is it takes a look at each processor and then breaks it down by core. Processors have cores, so its kind of processors within processors, and it takes each of those individual cores and manages them against the workload independently. So, if the workload doesn't require processor, it takes it down to a load near zero power state. So, you are not using power for something that is not happening, you know, the old way is the processor stayed ramped up regardless of the workload. Now, each core ramps down when there isn't workload associated with it. On the other hand, as you ramped down those other cores, they've got thermal (ph). In other words, the processor was designed to put out a certain amount of energy. Well, you're not using it all, that means that one core that you are using can be ramped up, overclocked, if you will, and so that means, a single threaded application that may be hitting this multicore processor can be maximized on that one core, and so you actually get extra performance of the one core that's being used and the cores that are not being used drop to zero. It's like a super transmission for a core that had multiple engines.

Karen Guglielmo: Okay, from what I know about intelligent power management solutions, it puts the processor and memory into the lowest available power state to meet the requirements of the current workload. Does this somehow affect performance?

Rob Enderle: Well, it can. What makes the Intel technology unique is the gap between low power and high power states is infinitesimal. I mean, otherwise, you would have this issue when a workload came in and required that power, you would have to ramp those cores back up again over a long period of time and you get this lag in performance. But because you are dealing with stuff down in silicon, it happens fast, and so, it's almost not noticeable from when it goes from lower power to high power state. So, it's almost like having your cake and eating it too. You get the energy efficiency of having a processor that's almost always in low power states until it's needed, but you get the performance of a processor that can operate at very high power performance levels depending on workload.

Karen Guglielmo: So, finally, for companies looking to implement an intelligent power management solution, how important is it to have a virtualized environment in place?

Rob Enderle: Well, a virtualized environment allows you to really manage the jobs against the server load as if each job had its own server and then to do that dynamically. So, as workloads come in, each one can be treated independently of the others. That gives you the greatest deal of flexibility and the greatest performance for a given server environment.

Karen Guglielmo: Okay, that about wraps up this segment on 'Strategies for Using Intelligent Power Management to Reduce Power Consumption in Downtime'. I'd like to again thank Rob Enderle for joining me today, and for more information on Intel's offering in this space, visit www.intel.com. Thank you again for joining us, and have a great day.