Home > Intel Communities > Open Port IT Community > IT@Intel > Data Center Blog > Blog > 2008 > September > 19
Up to Blog Posts in Data Center Blog
Previous Next

IT@Intel Data Center Blog

Currently Being Moderated
Don Atwood
8

 

Hi I’m Don Atwood, author of the newly released white paper and video that discusses our proof of concept (PoC) that tested cooling our Data Center with outside air. The topic of humidity control and if this would work in an ultra high humid climate keeps coming up. Most OEM spec’s allow for a wide range of humidity and it’s our belief that this cooling methodology could be used almost everywhere globally. Our only uncertainty comes around trying this near the ocean with high levels of salty corrosive wet air. We know it would negatively affect the servers at some point but the question is how quickly and is it within our refresh timetable. During a trip to ASIA last week I discussed trying a small scale “near the ocean” PoC to test this theory.. Does anyone thing this would add value to your company?

Tags: 45nm, consolidation, data_center, data_center_efficiency, data_center_power, datacenter, efficiency, energy, energy_efficiency, green, hpc, intel, intel_performance_libraries, intel_threading_building_blocks, it@intel, multicore, performance, power, server, server_consolidation, threading_analysis, virtualization, vtune, xeon, don, atwood


Add a comment Leave a comment on this blog post.
Sep 23, 2008 8:31 PM Don Atwood Don Atwood    says in response to Harold Simmons:

the video and white paper can been seen at:http://video.intel.com/?fr_story=2d6e0fbbef76b72c6119cc7fe7889bba20cb5192&rf=bm
Reply
Oct 1, 2008 2:12 AM Guest ciaran flanagan  says:

Don,

Great video and discussion on your RR experiment and its certainly creating a lot of buzz with my clients too...but a question :- the cost savings are the fresh air design relative to the traditional approach so how efficient ( in PUE ) is the traditional ? Knowing that will help clients really quantify the bet

 

cheers

 

Ciaran
Reply
Oct 9, 2008 1:45 PM Don Atwood Don Atwood    says in response to John Hazucha:

John – some good points.

 

On the failure rate - 2.45% (the traditional side of our PoC) is actually showing better than our enterprise high availability global DC’s (3.83%). The conditions within our enterprise DC is certainly tightly controlled and substantially more reliable and cleaner then the PoC but yet we still experienced a difference; we consider this a normal hard drive failure variance from Data Center to Data Center. With the anomaly in the numbers (the traditional PoC side out performing our Enterprise DC’s), we believe the failure difference is with well within normal failure rates we see globally and if we let it run for another 10 months the numbers will have shifted again. The primary failures we saw are 99% hard drive related and all fall within standard warranty replacement conditions (no additional labor for us). With 900 servers in one room, we swap failed drives every day.

 

What was not clarified in the white paper that you bring up is; if the increased failure means downtime to production app’s, we should consider this in the equation. Value vs. Risk - Our increase hardware failures translated to replacing a few more hard drives per week, nothing that actually caused an outage. We had 1800 hard drives spinning in the room.

 

When Intel looks at the “green” benefits of reducing total power consumption by more than 60% (or reduction of carbon footprint) and millions of gallons of water savings, Intel sees this as a big win, huge cost savings aside. Sure at the end of the day it’s a tremendous cost savings annually for us and our stockholders, but the combined cost savings and green benefit is a win-win for us.
Reply
Oct 10, 2008 11:06 AM Guest John Hazucha  says in response to Don Atwood:

Don: Thanks for the clarifications and additional info.

 

I am not familiar enough with the operational details at my client's site to know if drive failures cause them headaches or if drive replacement is just a normal part of their day, and does not cause any downtime. But, I can show them the white paper and your response, and let them make the decision.

 

If additional drive failures are a significant problem for them, they may be unwilling to take the risk that your assumption the failure rate falls within normal random variation is correct.

 

I was thinking that the other factor that may be significantly different between your enterprise DC and the PoC, and therefore account for the lower failure rate on the traditional side of the PoC, was the age of the hardware. Did you begin the experiment with new hardware in the PoC?
Reply
Dec 1, 2008 3:23 PM Guest Philip Gadd  says:

Don what speed was the network running at in this trial? Did you use optical or copper interconnects and have you considered using active copper cabling?

 

BR

 

Philip
Reply
Apr 9, 2009 8:02 AM Guest Graham Whitmore  says in response to Don Atwood:

Outside air free cooling has been around for many years with some apparent success stories.  The energy savings are beyond dispute, but the reason these systems are not widespread is the potential risk from airborne contaminants.  Humidification can be managed at a cost and filtration can be effective, except in the case of smoke, fumes or aerosols, which can be accidentally or intentionally introduced with devastating results.  These are the main stumbling blocks for mission critical designers, related to outside air.  Motivair has developed a 'hybrid' data center free cooling system that uses outside air but in a controlled and contained system.  The use of free cooling chillers allows the outside air to provide free cooling, but indirectly via additional 'free cooling' glycol coils mounted in front of the condenser coils inside the chiller.  This eliminates filtration, humidification & airborne contamination from the design equation.  There is a small sacrifice in efficiency compared to direct outside air cooling because of the additional heat exchange between outside air & glycol but the contamination risk is eliminated There are a number of installations around N. America currently in operation or in the implementation phase up to 6,500 kW.
Reply
Apr 15, 2009 7:59 PM Guest Kent Yoo  says:

Hi,

I'm working for one of IT infrastructure consruction companies in S. Korea.

After reading whitepaper of Air Economizer, I want to see more detailed and specific information such as cataloges or configurations.

I believe, with many opportunities, Air Economizer may be a killer solution for Data centers.

And it may be a big help for my company.

Please let me know more detailed information.

Thanks and regards,

Kent
Reply
Apr 28, 2009 9:07 AM Guest Rick Cockrell  says:

Don,  I know it's been a while since you did this study, but do you have any more information on this test case.  I am writing to find out if you still believe this is a sustainable pratice.  I am in the cooling industry and I have a system that runs from 6-12 kW of server to 1 kW of cooling while providing 57F SAT, 68F RAT and 45% RH +- 1.5% all year long. I use no outside air. As you can imagine, your report puts a damper on our sales.  Everyone is talking about outside air.  In your report you suggest that the server failure rate went from 2.43 to 4.4.  Is that a good thing?  I just ask because if my system failure rate went up that much in the warrenty period we'd be screwed.  Do you have any more information that I can show a customer on the failure rates after the 8 months?  If you do, please contact me at the e-mail I provided it will help me alot.  Our system uses refrigerant side economizing and usually provides a 1-4 Year ROI depending on the cost of energy and drops right in to replace CRAC's.  I am a young inventor, I really need your help.  The world also needs your help with the e-waste issue. ASE is not a sustainable pratice if your failure really go up 100%.
Reply

Actions

Popular Blog Posts