[AusNOG] Internode goes Carbon Neutral

[Shaun Dwyer]

In the not too distant past, I was looking at some APC hot-aisle containment DC solutions.

An example DC they took me on a tour of had no raised floor, two rows of racks, fully contained, in-row chillers that sucked hot air out of the completely contained hot area and chilled off the larger volume of the room. All cabling (power and data) are done from above. One of the obvious benefits of hot-aisle containment vs cold-aisle is that hot-aisle means you have a larger reserve volume of cold air in the event of failure.

They claim huge energy efficiencies as compared to conventional raised floor cooling. They don't compare it to a standard raised, underfloor chilled system with any of the suggested improvements.

APC block off unpopulated areas of the racks to attempt to force the airflow through the servers and give even cooling.

All sounds good in theory, but they do charge a pretty penny - worth a look, even if its just to get some ideas on how someone else is doing it.

After seeing APC's example DC, I'm sure in a hot aisle containment system it wouldn't be too hard to add a few sensors, a bit of ducting to the outside, a fan, a method to seal it all up when not evacuating, and some microcontroller love. Infact, I'm sure its something that would be extremely easy to put together with an AVR.

One obvious pitfall that comes to mind though is making sure you don't cause a flood in your DC from all the moisture that gets condensed out of the colder outside air thats coming into the DC to replace your hot exhaust.

In Australia, this sort of thing could only be used as a method of saving electricity on cooling; you'd still need enough cooling capacity to handle your full thermal load in the event that the atmosphere outside is hotter than your contained hot air.


Cheers!
-Shaun


On 19/11/2009, at 8:05 AM, Bill Walker wrote:

> If you contain the Cold aisle and have a common hot area, then you can vent the hot into space. IMO a free air chiller is a better option. But retro fitting a vent is much easier than a new chiller.
> 
> Most of the major rack vendors can do in row cooling / cold aisle containment solutions, I've seen solutions from:
> 
> APC
> Rittal
> And Rack Technologies
> 
> The Rittal one stands out as their coolers intake only requires the water to be at 15 degree, which at least in NZ means we can use tap water in the event of a chiller failure. But Rack technologies will give you the option of a fully tailored rack.
> 
> Cheers,
> 
> Bill
> 
> -----Original Message-----
> From: ausnog-bounces at lists.ausnog.net [mailto:ausnog-bounces at lists.ausnog.net] On Behalf Of Matt Carter
> Sent: Thursday, 19 November 2009 12:42 p.m.
> To: 'Curtis Bayne'; Mark Prior
> Cc: ausnog at ausnog.net
> Subject: Re: [AusNOG] Internode goes Carbon Neutral
> 
>> 
>> What about venting hot air if atmosphere temperatue is lower than the
>> outtake of the hot aisle (especially if our hot aisle is contained).
> 
> Who here is actually doing this in .au ? I know of a variety of DC in the New York area doing this, curious to know who has or is in the process of building DC's that have these innovations. It's something our new DC will be doing from day 1, would certainly be nice to chat to some people who are also doing it and have experienced the challenges of the Australian environment.
> 
>> Datacentre efficency doesn't have to be hard or expensive, most of the
>> time it is a combination of common sense and pragmatic implimentation.
> 
> One of the very interesting things I am seeing is what appears to be an apparent use of other technologies to overcome outright design efficiencies (which then exclude other future options of a traditional space) Eg in-row and in-rack cooling has become rather popular in certain areas, and although I don't disagree it has it's applications, yep, sure, you have a container to work with, your ceilings are 20 ft high, you've been given a tin shed, etc. BUT, if you have the capacity to build the DC from scratch and you have the capacity to design it properly, with good hot/cold separation, overhead plenums, etc, and it's going to be around for a while, the assertion that things like in-row are better than a properly designed DC, are imho, a fallacy. Esp when you consider things like the option to vent the hot aisle to the outside, which I don't see how you could do easily with in-rack or in-row cooling, (but I'm not a expert in this field either).
> 
> <non colo provider related stuff skipped>
> 
>> I would be very interested to hear from anyone who is employing of
>> these optimization strategies in their organization.
> 
> As you mention a lot of this is common sense from the lessons others have learned over many years. Eg today still, DC being built where the practice is still to use perimeter cooling design, and it's just left like that. Is there any containment between hot and cold? No. is there any exhaust chimney into overhead plenum in the roof space? No. Have we even made any attempt to adjust the grills so we have even at least even distribution of air through the rack instead of 10x as much CFM at the ends versus the middle? No. If we have a bit of a mish mash and poor hot/aisle containment, have we attempted to do any CFD modeling and/or reorganise? Then this DC be shocked and horrified when <insert vendor here> comes along and shows the massive savings they can make by moving to another cooling methodology!! That said, I do appreciate, we all do what we can with the budgets we have, and that has a factor to play, but if we are trying to claim a 'state of the art' facility ...........
> 
> --matt
> 
> _______________________________________________
> AusNOG mailing list
> AusNOG at lists.ausnog.net
> http://lists.ausnog.net/mailman/listinfo/ausnog
> 
> No virus found in this incoming message.
> Checked by AVG - www.avg.com 
> Version: 8.5.425 / Virus Database: 270.14.72/2511 - Release Date: 11/18/09 07:50:00
> _______________________________________________
> AusNOG mailing list
> AusNOG at lists.ausnog.net
> http://lists.ausnog.net/mailman/listinfo/ausnog