• Add a Comment
  • Edit
  • More Actions v
  • Quarantine this Entry

Comments (2)

1 Bill Smart commented Permalink

I think you may have missed on a primary reason. I've known a few IT Managers and I can say that at least a few of them have had a fair sized ego :-) If they could take action to reduce the amount of energy consumed in the data center and document the savings in real cash, then they can raise their stature in the organization. Everybody wants to be a hero, especially the guy who occasionally gets beat up... it helps even the score. Bill.

2 Todd Gatts commented Permalink

I think it's useful to think of the energy monitoring as divided into energy management and risk monitoring. Energy management is all about efficiency; risk monitoring is all about avoidance.

 
 
We monitor energy use in order to know where we are in the energy efficiency landscape. In the beginning, our measurements are a stake in the ground. We know we were "here" when we started. We also use "here" to compare ourselves to "there". "There" could be the energy measurements of our other data centers, or data centers run by other companies in our industry, or what the Green Grid says we should be striving for.
 
 
So the answer the question, "Why should the IT Manager care?" is quite simple. Because someone above him is looking at here and looking at there, and asking the IT Manager why they aren't any closer, and what is the plan to bring them closer, and what is the improvement goal for the next year?
 
 
And so the IT Manager starts to make changes to move from here to there. And after each change the IT Manager takes new measurements, and decides if the move had the correct effect: Did the change reduce energy consumption of the data center without affecting business service performance?
 
 
Energy monitoring gives you quite a few tools to understand where the energy is being consumed. Energy is consumed directly by IT equipment -- servers, networks, storage,etc -- but energy is also consumed by the systems which support the IT equipment. The two most relevant systems supporting IT equipment are power and cooling. And as soon as you say cooling, you realize that it is also important to monitor the production of heat, since it will take energy to get the heat out of the building through the cooling system.
 
 
The IT Manager can use the power and thermal monitoring to gain visibility into the condition of the data center and can make sensible decisions about what to change and by how much. "Can we run the data center warmer?" Monitor how evenly the cool air is being supplied to the servers. "Are my computer room air conditioners (CRAC) sharing the load properly?" Monitor the cooling output of each CRAC and see if any are always working harder than the rest.
 
 
The IT Manager removes a perforated tile here and unblocks a cable chase there and now the data center cooling is much more even, and the cooling system set-point can be raised two degrees without affecting the servers. "Here" just moved closer to "there".
 
 
One of the reasons that there are a lot of things an IT Manager can do to save energy is because data centers are overbuilt. If you don't know if there are any cooling problems in the data center, what do you do? You make it too cold in the data center so there is some cooling buffer to handle the servers that are not being cooled evenly. If you don't know how much power a data center will draw at peak demand, what do you do? You build the power distribution system extra large so that it can handle a theoretical maximum.
 
 
Energy monitoring lets you see the excesses in your data center and lets you squeeze out the excesses with confidence, helping you create a lean data center. A lean data center is a good thing, but a lean data center has a unique set of problems as well. Overbuilt data centers may have too much buffer, but a lean data center may have very little. When problems occur in an overbuilt data center, the buffer may absorb the effect without anyone knowing. When problems occur in a lean data center, action needs to be taken quickly. And this is where energy management leads to risk monitoring.
 
 
There is a funny dichotomy in the infrastructure management systems which support a data center. The behavior they are designed to do, is, in some ways, the most dangerous thing they could do. These management systems are designed to automatically react to and recover from negative events occurring outside the data center. One measure of how well a management system has done its job handling an event is how little the IT equipment notices the event.
 
 
Loss of utility power? No problem, the uninterruptible power supplies (UPS) will pick up the power load in a fraction of a power wavelength ... the network routers never heard a thing. Loss of a chiller? No problem, the stored chilled water can supply the CRACs for 30 minutes ... the servers are as comfortable as ever.
 
 
So it's a good thing that the infrastructure hides events from the IT equipment, but it's a bad thing when it hides the event from the IT Manager. Now that the buffer has been squeezed out of the system, the data center doesn't have excess capacity. You need to know that all your CRACs are cooling at 92% of capacity and it's only 11 in the morning. You need to know you have branch circuits that have exceeded 95% of capacity and stock trading starts in a hour. You need to know when there is a city-wide power brownout and your UPSes are handling a quarter of the data center power demand. When something happens that places the data center under added risk, the IT Manager needs to know about it right away.
 
 
There are lots of things an IT Manager can do if the problem is apparent: move workload to a different data center, load your most efficient servers to their maximum and shut down your least efficient servers, delay maintenance jobs, etc. So the answer the question, "Why should the IT Manager care?" is quite simple. Having a disaster does not make an IT Manager look good; handling or avoiding a disaster does.
 
 
Energy monitoring progresses down two related paths, energy management and risk monitoring. You may start down the path of energy management -- taking measurements, comparing your measurements to industry standards, making changes -- but that path inevitably branches onto the path of risk monitoring -- knowing how close you are to your power and cooling limits, knowing when an infrastructure event has occurred. The IT Manager has to walk both of these paths, and that's why the IT Manager cares.
 
 
Todd

Add a Comment Add a Comment