stand by computers in server room
-
Remember, two killers of electronics: moisture and heat.
-
when i access server room, the metal looks wet, is this show that there is humidity right?
-
@IT-ADMIN said:
when i access server room, the metal looks wet, is this show that there is humidity right?
You really need to get some humidity control and air conditioning (not just cooling) in there. If the metal looks wet chances are it is wet.
-
@IT-ADMIN said:
when i access server room, the metal looks wet, is this show that there is humidity right?
Yup. And remember, even if it isn't full-on droplets of water, if you're getting enough moisture that it's starting to actually do to your motherboards what it's doing to the door, you have a SERIOUS problem. I'd start by at least putting some dehumidifiers in there until a more permanent solution can be arranged. Just bear in mind, those have to be emptied usually.
-
@IT-ADMIN said:
when i access server room, the metal looks wet, is this show that there is humidity right?
Yeah. It should NEVER look wet. You need a hygrometer in there, always. You should not be guessing, you need to KNOW what the humidity is at all times, even more than the temperature (which you would feel if it got too hot.)
-
Why is your server room at 17 anyway? That is far, far below recommended temperatures.
-
@scottalanmiller said:
Why is your server room at 17 anyway? That is far, far below recommended temperatures.
And if you're dealing with high humidity as it is, this will only amplify the condensation problem by keeping the temperature lower...
-
@thanksajdotcom said:
@scottalanmiller said:
Why is your server room at 17 anyway? That is far, far below recommended temperatures.
And if you're dealing with high humidity as it is, this will only amplify the condensation problem by keeping the temperature lower...
Yup, I'd look at raising that at least to 25.
-
Intel and IBM tests in similar weather conditions (New Mexico instead of Qatar) years ago put optimum efficiency of the datacenter closer to 30-32.
-
@scottalanmiller said:
Intel and IBM tests in similar weather conditions (New Mexico instead of Qatar) years ago put optimum efficiency of the datacenter closer to 30-32.
Wow, that's 86-90F. That seems very warm!
-
@scottalanmiller said:
@thanksajdotcom said:
@scottalanmiller said:
Why is your server room at 17 anyway? That is far, far below recommended temperatures.
And if you're dealing with high humidity as it is, this will only amplify the condensation problem by keeping the temperature lower...
Yup, I'd look at raising that at least to 25.
the problem is the high temperature we have in qatar, usually 46 - 38 C, so if i raise the temperature, i risk to kill the servers
-
@thanksajdotcom said:
@scottalanmiller said:
Intel and IBM tests in similar weather conditions (New Mexico instead of Qatar) years ago put optimum efficiency of the datacenter closer to 30-32.
Wow, that's 86-90F. That seems very warm!
We studied that project in grad school. It was pretty cool how they set it up. As long as humidity didn't get too high you could have ambient temperatures of ~100F before things started to fail.
-
@IT-ADMIN said:
@scottalanmiller said:
@thanksajdotcom said:
@scottalanmiller said:
Why is your server room at 17 anyway? That is far, far below recommended temperatures.
And if you're dealing with high humidity as it is, this will only amplify the condensation problem by keeping the temperature lower...
Yup, I'd look at raising that at least to 25.
the problem is the high temperature we have in qatar, usually 46 - 38 C, so if i raise the temperature, i risk to kill the servers
Your server room should have independent air conditioning. Keeping it around 20-25 C should be fine.
-
@coliver said:
@thanksajdotcom said:
@scottalanmiller said:
Intel and IBM tests in similar weather conditions (New Mexico instead of Qatar) years ago put optimum efficiency of the datacenter closer to 30-32.
Wow, that's 86-90F. That seems very warm!
We studied that project in grad school. It was pretty cool how they set it up. As long as humidity didn't get too high you could have ambient temperatures of ~100F before things started to fail.
I guess that makes sense. I mean, as a mostly-human being, Syracuse during the summer is often infinitely harder to handle than Dallas ever was due to how humidity affects perception of temperature. I guess it wouldn't be THAT different to electronics.
-
@thanksajdotcom said:
Wow, that's 86-90F. That seems very warm!
It sure does, because IT has this cultural obsession with thinking only in terms of the servers themselves and not holistically for the business. It's "IT at any cost" thinking. That and poor airflow forces datacenters to go with super cold air to make up for heated air pockets.
But servers do not need the cold air that IT generally assumes. That's a myth that just keeps getting repeated. Good airflow and holistic business thinking put the optimum temperatures much higher in cases where the cold air is not essentially free (like Scotland.)
-
@coliver said:
@IT-ADMIN said:
@scottalanmiller said:
@thanksajdotcom said:
@scottalanmiller said:
Why is your server room at 17 anyway? That is far, far below recommended temperatures.
And if you're dealing with high humidity as it is, this will only amplify the condensation problem by keeping the temperature lower...
Yup, I'd look at raising that at least to 25.
the problem is the high temperature we have in qatar, usually 46 - 38 C, so if i raise the temperature, i risk to kill the servers
Your server room should have independent air conditioning. Keeping it around 20-25 C should be fine.
What matters is keeping it consistent, not the temp itself.
-
@IT-ADMIN said:
@scottalanmiller said:
@thanksajdotcom said:
@scottalanmiller said:
Why is your server room at 17 anyway? That is far, far below recommended temperatures.
And if you're dealing with high humidity as it is, this will only amplify the condensation problem by keeping the temperature lower...
Yup, I'd look at raising that at least to 25.
the problem is the high temperature we have in qatar, usually 46 - 38 C, so if i raise the temperature, i risk to kill the servers
That's NOT how air conditioning works. The temperature outside has nothing to do with the temperature inside. If you set it to 17, it is 17 no matter what the temp is outside. If you set it to 25, it is the same 25 as for any of us. Your 25 is not hotter than ours. What DOES matter is that the cost of cooling from 46 to 17 is much, much higher than cooling from 25 to 17. So the price for going unnecessarily low is more for you.
-
@coliver said:
@thanksajdotcom said:
@scottalanmiller said:
Intel and IBM tests in similar weather conditions (New Mexico instead of Qatar) years ago put optimum efficiency of the datacenter closer to 30-32.
Wow, that's 86-90F. That seems very warm!
We studied that project in grad school. It was pretty cool how they set it up. As long as humidity didn't get too high you could have ambient temperatures of ~100F before things started to fail.
We've done these tests (nonvoluntarily) with HP servers and they ran happily towards 160F!! But I would not recommend it since humans can't go in the room to replace parts.
-
@thanksajdotcom said:
@coliver said:
@thanksajdotcom said:
@scottalanmiller said:
Intel and IBM tests in similar weather conditions (New Mexico instead of Qatar) years ago put optimum efficiency of the datacenter closer to 30-32.
Wow, that's 86-90F. That seems very warm!
We studied that project in grad school. It was pretty cool how they set it up. As long as humidity didn't get too high you could have ambient temperatures of ~100F before things started to fail.
I guess that makes sense. I mean, as a mostly-human being, Syracuse during the summer is often infinitely harder to handle than Dallas ever was due to how humidity affects perception of temperature. I guess it wouldn't be THAT different to electronics.
Although it is opposite, for electronics which are air cooled, high humidity increases cooling potential. For humans which are evaporation cooled, humidity decreases cooling potential.
-
IBM has shown that running at 37C consistently is far better for your gear than bouncing around between 15C and 20C.
