NOREX News section will be updated frequently with member alerts, news releases, helpful links, event notices, member gold nuggets, resource additions and helpful community building information. If it's new, you will find it here.
If you keep the temperature in your data center below 75 degrees, you may be wasting money. That according to a engineer who participated in our recent Data Center Efficiencies WebForum.
Topic: Optimal data center temperature
Moderator: I put together a poll based on Markâ€™s comment that 75 degrees is the optimal temperature.
Moderator: It looks like some of the participants are keeping it under 70 degrees, the majority between 71-75 degrees. Mark, are you saying that it is unnecessary to be below 70 degrees, or does that depend?
Mark S.: Actually there is new documentation that says 80 degrees is fine. We have received from our customers a consistent message that 70 degrees, well, there isnâ€™t anything magical about it anymore. I donâ€™t even know that there ever has been.
As an engineer when I look at it all aspects of your operating margin, when you are routing in your room at 68 degrees you have got some time margin if things go wrong. You have got time margins built into that temperature that will help you deal with any issues that come up. As we decrease, as we raise that temperature we do affect those margins. There is no question about it, but it is at a cost, or at a savings in this case, so that is why it is being pursued.
Steve K.: I have one question, if I recall correctly the discussion was you can have up to 80 degrees in the data center?
Mark S.: We are seeing customers approaching 80 on a regular basis. We are setting our systems up at about 75 degrees but that can be adjusted quite readily.
Steve K.: Do you know if that is based upon servers only? We have SAN units in ours and we have had warnings go off from the SAN units that I think at 74-75 degrees the SAN units start complaining about the temperature.
Mark S.: Is the unit getting that kind of temperature in its front? Usually what happens is these servers are instrumented for temperature but they are instrumented inside of the server so they are tied into heat sinks inside.
So the first question would be are your SANs getting enough air flow to meet their needs at that temperature or are they in a situation where we are using under the floor air, where we are trying to blow it and we are perhaps not getting enough and we are perhaps sucking some hot air across the top of the rack to supplement the air supply.
Those are the types of things that we see. If that is the case it would be evidenced by the fact that the units in the top of the server are getting much warmer than the ones on the bottom. You might check for that to see if it is happening. That indicates a lack of sufficient air supply.
Jim C.: One of the best investments that I made recently is a little laser temperature gun that you can point at the different spots on a server rack and you can actually see the different temperatures, how the temperature might range. Typically your thermostats on your air conditioning units are four feet off the floor. There is a wide temperature change between floor level and the top of the racks.
Moderator: Have you made any design changes or moved anything around configuration-wise as a result?
Jim C.: Yes, we have. We are under floor for our cooling and we have relocated the perforated floor tiles in some instances to make more airflow available. We have closed off others as a result of that. Very early on we addressed all of the hot aisle/cold aisle issues but if you want to make sure that you are getting the right temperature at the right spot you want to monitor the temperature where your problem areas are. If that is a critical resource then that is where you want to put your effort into it.
Moderator: Jim, what was the temperature that you found using the laser gun, and then what did you set it at after making the change? Do you recall?
Jim C.: I donâ€™t recall. It has been a couple of years back already but you definitely need to be paying attention to temperature coming into rack space.
Mark S.: One thing that I would say is that when we set up these systems we look at the power going into the rack and the temperature going in and out. Those are critical instrumentation points so I can see how the system is operating because knowing the power and knowing the temperature going in I can calculate what the temperature going out is going to be so it can give me some cross check references.