Are my BTU figures for my (small) machine room crazy?

>Greetings,
>I'm trying to figure out how much cooling capacity we need
>for our small machine room. The numbers I'm getting seem
>absurdly high, so I thought I'd run it past the denizens of
>this newsgroup and see if anyone has any suggestions.
>The room is 9 ft. x 17 ft. x 10 ft. high. There's no drop
>ceiling or raised floor. Two vents from the house A/C
>discharge into the room, and there's a return vent venting
>into the drop ceiling in the hallway. It's not a designed
>machine room per se; it's more of a big closet that we're
>using to hold our racks.
>There are 4 long flourescent bulbs in the room, which I
>suspect don't add much to the heat load.
>We've got two racks of hardware, a couple of small desktops
>and a CRT monitor. One of the racks is about 2/3 full of
>various pieces of network hardware -- PoE switches, switches,
>minihubs, CSU/DSU, and a couple of breadbox PCs that are our
>firewalls.
>The other rack is pretty much full of servers, most of which
>are 1U servers whose power load is 560W and whose specs claim
>a BTU rating of 2750 (which seems a bit high, but that's what
>the specs say).
>When we were relying on just the house A/C vents to cool the
>room, it was always very warm in front of the racks, and
>downright hot in back of them. When the temperature behind
>the racks hit 100 deg. F and the HDD temperature monitor I
>was running on one of the desktops hit 54 deg. C, I decided
>it was time to do something about the problem. I called Spot
>Coolers and described the situation, and they suggested a
>1.5-ton MovinCool unit, which we agreed to. We ran 12-inch
>flexible plastic ducting from the cooler to the return vent.
>The unit managed to get the temperature in the room
>(according to its thermometer) down to around 74 degrees, and
>the temperature behind the racks got down to around 80. Then
>we turned on a few more servers, and now the room temperature
>is hovering around 79 and the behind the racks it's up to
>around 90. I find it hard to believe that the few servers we
>turned on made that big of a difference, so I suspect that
>there are other environmental differences going on, although
>I'm not sure what they could be.
>Since we would ideally like to keep the room around 68
>degrees, clearly the 1.5-ton unit isn't big enough. The
>question is, how much capacity do we need?
>I took an inventory of all the hardware in the room and
>recorded from each item's specifications the BTU if given or
>the wattage * 3.5 (a metric I found in several places on-line)
>otherwise. When I totaled everything, I got 75,103 BTU, or
>6.3 tons. I find it extremely hard to believe that less than
>two full racks of hardware can generate that much heat, but
>perhaps I'm hearkening back in the days when most servers only
>had one CPU :-).
>I guess the question I need sanity-checked by the experts here
>is, can two racks of hardware reasonably generate that much
>heat? If so, and we want to leave open the possibility of
>adding a third rack of hardware to the room, it seems to me
>that we're going to have to get a 10-ton unit. Are there even
>10-ton coolers that'll fit in a small machine room?
>Thanks for any help and advice you can provide.

>Greetings,
>I'm trying to figure out how much cooling capacity we need
>for our small machine room. The numbers I'm getting seem
>absurdly high, so I thought I'd run it past the denizens of
>this newsgroup and see if anyone has any suggestions.
>The room is 9 ft. x 17 ft. x 10 ft. high. There's no drop
>ceiling or raised floor. Two vents from the house A/C
>discharge into the room, and there's a return vent venting
>into the drop ceiling in the hallway. It's not a designed
>machine room per se; it's more of a big closet that we're
>using to hold our racks.
>There are 4 long flourescent bulbs in the room, which I
>suspect don't add much to the heat load.
>We've got two racks of hardware, a couple of small desktops
>and a CRT monitor. One of the racks is about 2/3 full of
>various pieces of network hardware -- PoE switches, switches,
>minihubs, CSU/DSU, and a couple of breadbox PCs that are our
>firewalls.
>The other rack is pretty much full of servers, most of which
>are 1U servers whose power load is 560W and whose specs claim
>a BTU rating of 2750 (which seems a bit high, but that's what
>the specs say).
>When we were relying on just the house A/C vents to cool the
>room, it was always very warm in front of the racks, and
>downright hot in back of them. When the temperature behind
>the racks hit 100 deg. F and the HDD temperature monitor I
>was running on one of the desktops hit 54 deg. C, I decided
>it was time to do something about the problem. I called Spot
>Coolers and described the situation, and they suggested a
>1.5-ton MovinCool unit, which we agreed to. We ran 12-inch
>flexible plastic ducting from the cooler to the return vent.
>The unit managed to get the temperature in the room
>(according to its thermometer) down to around 74 degrees, and
>the temperature behind the racks got down to around 80. Then
>we turned on a few more servers, and now the room temperature
>is hovering around 79 and the behind the racks it's up to
>around 90. I find it hard to believe that the few servers we
>turned on made that big of a difference, so I suspect that
>there are other environmental differences going on, although
>I'm not sure what they could be.
>Since we would ideally like to keep the room around 68
>degrees, clearly the 1.5-ton unit isn't big enough. The
>question is, how much capacity do we need?
>I took an inventory of all the hardware in the room and
>recorded from each item's specifications the BTU if given or
>the wattage * 3.5 (a metric I found in several places on-line)
>otherwise. When I totaled everything, I got 75,103 BTU, or
>6.3 tons. I find it extremely hard to believe that less than
>two full racks of hardware can generate that much heat, but
>perhaps I'm hearkening back in the days when most servers only
>had one CPU :-).
>I guess the question I need sanity-checked by the experts here
>is, can two racks of hardware reasonably generate that much
>heat? If so, and we want to leave open the possibility of
>adding a third rack of hardware to the room, it seems to me
>that we're going to have to get a 10-ton unit. Are there even
>10-ton coolers that'll fit in a small machine room?
>Thanks for any help and advice you can provide.

> jik@kamens.brookline.ma.us (Jonathan Kamens) wrote:

>>I guess the question I need sanity-checked by the experts here
>>is, can two racks of hardware reasonably generate that much
>>heat? If so, and we want to leave open the possibility of
>>adding a third rack of hardware to the room, it seems to me
>>that we're going to have to get a 10-ton unit. Are there even
>>10-ton coolers that'll fit in a small machine room?

>Although you've did a fair job of describing your particular
>situation, there are still far too many unknowns to provide you with
>the answers you seek.

>Is it terribly brilliant to have all of your cooling capacity in one
>unit, so that when it goes down, even for routine maintenance, your
>servers go down as well?

>Install a 2 - 3 ton ductless mini-split with inverter technology.
>Most come with low-ambient cooling as standard features.
>As load increases, add another ductless mini split. Multiple units
>with overlapping capacity is far more desirable than one stand-alone.

>And yes, Liebert has a 10 ton, 2 compressor model that will easily fit
>in your room. Of course you will need to locate the remote condensor
>somewhere else, either roof or side yard.

>Do you even have the required additional ampacity in your electrical
>panel for a 10 ton unit?

> My biggest
> question at this point is, is it really possible for less than
> two racks of hardware to produce over 6 tons of heating
> capacity,

> >Do you even have the required additional ampacity in your electrical
> >panel for a 10 ton unit?

> If not, then we'll just have to install new circuits.