Comments on: Chiller-Side Chats: The Capacity Problem http://knowledge.digitalrealtytrust.com/2009/07/chiller-side-chats-the-capacity-problem/ Wed, 16 Dec 2009 12:33:31 +0000 hourly 1 http://wordpress.org/?v=3.0.1 By: Mike Manos http://knowledge.digitalrealtytrust.com/2009/07/chiller-side-chats-the-capacity-problem/comment-page-1/#comment-36 Mike Manos Tue, 15 Sep 2009 09:08:49 +0000 http://knowledge.digitalrealtytrust.com/2009/07/chiller-side-chats-the-capacity-problem/#comment-36 James - Great points. I think the unfortunately the IT world is probably not paying attention to this at all. It further drives the requirement for these organizations to work closely with one another as the problem space is getting very complicated. To add to that, I do not believe Power Capping is ready just yet for prime time (see my other post on power capping). James – Great points. I think the unfortunately the IT world is probably not paying attention to this at all. It further drives the requirement for these organizations to work closely with one another as the problem space is getting very complicated. To add to that, I do not believe Power Capping is ready just yet for prime time (see my other post on power capping).

]]> By: James Longacre http://knowledge.digitalrealtytrust.com/2009/07/chiller-side-chats-the-capacity-problem/comment-page-1/#comment-35 James Longacre Mon, 14 Sep 2009 20:31:19 +0000 http://knowledge.digitalrealtytrust.com/2009/07/chiller-side-chats-the-capacity-problem/#comment-35 I recently had the experience of the impact of the new NEC 2008 code changes. UPS sizing has now been set to the atteched rated load and not the running or known load of a IT system. The ultimate problem that presents itself is how do you explain to the customer that he is paying for a 400Kw UPs and in reality he is only going to use 230Kw of it? Compounding this is the issue of the circuit balance and the harmonics caused from having way too much transformer and never loading it, big cables and never using them and the cost to build out to the rated power of the servers and never reach the load. How do we as electrical folks remain in compliance and not have a poorly performing electrical system because of all or mostly non linear load? Sure we can add devices to offset the harmonics, but that only works for one load range and the cost of active cancelling is in my view- "unreasonable". So my question to you is "How is the IT industry dealing with this from a code point of view and a operational point of view or are they doing anything at all?". This forces the need for power capping, but as you say, this is not a well understood feature at the server level. It further confounds the issue of acceptable control methods and traditionallly digital communications to regulate power has not been well recieved by the power engineers, they want hard points and settings to limit power, not server scripting. So where does all this start to be addressed? What is leading the design of the facility? so many questions and so few answers... James I recently had the experience of the impact of the new NEC 2008 code changes.
UPS sizing has now been set to the atteched rated load and not the running or known load of a IT system. The ultimate problem that presents itself is how do you explain to the customer that he is paying for a 400Kw UPs and in reality he is only going to use 230Kw of it? Compounding this is the issue of the circuit balance and the harmonics caused from having way too much transformer and never loading it, big cables and never using them and the cost to build out to the rated power of the servers and never reach the load. How do we as electrical folks remain in compliance and not have a poorly performing electrical system because of all or mostly non linear load? Sure we can add devices to offset the harmonics, but that only works for one load range and the cost of active cancelling is in my view- “unreasonable”.
So my question to you is “How is the IT industry dealing with this from a code point of view and a operational point of view or are they doing anything at all?”.
This forces the need for power capping, but as you say, this is not a well understood feature at the server level.
It further confounds the issue of acceptable control methods and traditionallly digital communications to regulate power has not been well recieved by the power engineers, they want hard points and settings to limit power, not server scripting. So where does all this start to be addressed? What is leading the design of the facility? so many questions and so few answers…

James

]]>