Found an interesting blog entry by Philip Evans.

Phil Evans is an Infrastructure Optimisation consultant with Microsoft Services in the UK. He is a passionate believer in right-size computing and helps organisations get the best out of their IT infrastructure spend.

Phil uses a simple concept of equating IT equipment to electric space heaters running 24 x 7 in a refrigerator.

It's that simple, but few organisations are prepared to tackle poor utilisation of their server infrastructure.  In the economic boom years, running an empty fridge was not a big deal.  Today, however, it's different.  Take Internet Service Providers - do they run servers that do little or nothing?  I think not - they will provision new hardware when they absolutely have to and not before.  Doing so is exactly how they steer a course between operating costs and profits.

The average server power supply runs at about 800 Watts peak capacity.  That's having a home electric heater or vacuum cleaner on 24/7, producing lots of carbon.  Now imagine the 300 vacuum cleaners being stored in a large fridge 24/7 and that's what you have in the average data center.  It just doesn't add up.

Storage Area Networks (SANS)are even worse.  These are intensively-packed units of hard disk drives that spin constantly, even when not in use.  SAN units (aggregated into large collections) typically use 1500-2000 Watts EACH, meaning that the average implementation could use up to 40 Kilowatts of power - that's 40 electric fires (in a large fridge, don't forget) running 24/7.

Then consider the fridge itself.  Would you put a heater in a fridge?  Of course not - but that is exactly what a data center is and the only means of cooling for most installations is to chill air and push that air in.  Doing so actually requires more energy than is actually being generated within the data center (simple laws of physics come into play here).  Wouldn't it be better to just regulate the temperature of the incoming air into a data center rather than trying to isolate it from an energy perspective?

We take it for granted PUE is important, but many users don’t get it that data centers are like refrigerators full of space heaters. You can continue the story by saying a typical refrigerator cycles on and off and has a life cycle of 14 years, but imagine the stress put on the refrigerator as it runs 24 x 7 cooling the space heaters.  The lifespan is now less than 5 years.  Then you need battery backup and generators to keep your space heaters and refrigerators running when the power goes out.

Others may understand how their servers impact the data center infrastructure using this story.  I am going to try this in a presentation. 

Thanks Philip for putting your idea out there.

Next week is Uptime Symposium.  Rich Miller will be there as well for Data Center Knowledge, and we are both on the same panel.

The Uptime Institute’s four-day Symposium 2009 runs next Monday through Thursday (April 13-16) at the Hilton New York. Day passes are available. In a late addition, Google’s Chris Malone will be on hand to discuss the company’s data center innovation and energy efficiency. “Google is using industry-proven concepts in radically different ways, which has led to millions of dollars in savings for the company,” said Kenneth G. Brill, executive director of the Uptime Institute. “By making appropriate business tradeoffs that optimize reliability, performance, and capital and operational expenditures, the Institute estimates that Google has saved at least $500 million in capital expenditures alone. The point here is that any company with a similar business case, application portfolio and appetite for risk can accomplish equally striking results.” Malone will present on Tuesday at noon. NOTE: I’ll be participating in a panel Tuesday at 3:30 pm on “Sustainable Computing in the News.” If you’re at the conference, stop by to say hello.

Last year, I blogged on Mike Manos’s opening keynote and it was a good one to set the tone for following year.

Microsoft's Mike Manos Opening Keynote Uptime Institute, Green Enterprise Computing

The opening of the Uptime Institute Symposium started with Pitt Turner, saying "what are we doing here?" Green Enterprise Computing is a timely topic for the industry event. I've seen Mike present before, and here is what I got out of Mike's latest presentation as new information to digest.

1. Mike's call to action for all was to stop being information hogs, and to share with the industry. He gave numerous example, and here are a few Mike shared.
2. Mike emphasized that in spite of a focus on technology. Microsoft has found having the right people and processes makes bigger impact than technology . Mike states over 50% of data center outages are caused by human error. This contrasts a common method to invest in multiple layers of infrastructure redundancy to achieve uptime.Mike was proud of Microsoft's ability to have a 100% facilities uptime over 7 years by implementing strong and disciplined maintenance programs. Also, redundant infrastructure creates more energy waste.
3. Technology is not the only answer to energy efficiency. People are the opportunity. Microsoft achieved a PUE improvement from 2.2 to 1.8 with no new technology just by people making changes to existing systems in one of their older facilities.
4. Mike didn't say this, but bottom line he emphasizes an Amazon.com approach in getting data on everything, and they'll figure out what to do with it after by giving people the information to do the right thing.