Data Center Power Efficiency

James Hamilton is one of the smartest and most accomplished engineers I know. He now leads Microsoft’s Data Center Futures Team, and has been pushing the opportunities in data center efficiency and internet scale services both inside & outside Microsoft. His most recent post explores misconceptions about the Cost of Power in Large-Scale Data Centers:


I’m not sure how many times I’ve read or been told that power is the number one cost in a modern mega-data center, but it has been a frequent refrain. And, like many stories that get told and retold, there is an element of truth to the it. Power is absolutely the fastest growing operational costs of a high-scale service. Except for server hardware costs, power and costs functionally related to power usually do dominate.

However, it turns out that power alone itself isn’t anywhere close to the most significant a cost. Let’s look at this more deeply. If you amortize power distribution and cooling systems infrastructure over 15 years and amortize server costs over 3 years, you can get a fair comparative picture of how server costs compare to infrastructure (power distribution and cooling). But how to compare the capital costs of server, and power and cooling infrastructure with that monthly bill for power?

The approach I took is to convert everything into a monthly charge. […]

James Hamilton explains Datacenter Costs
tags: , , , , , , , , , ,
  • It’s always good to keep these statements on actual vs perceived energy consumption in check… thanks for the post.

    Where I’d like to see more exploration about the is not simply how Data centers can be more efficient- but more resilient.

    Energy storage seems to be the biggest thing – not on the radar of most people in the world. Batteries, fuel cells and capacitors are ripe for integration into the grid– and data centers could prove to be nice early adopter markets.

    Just a thought..

    Garry Golden

  • Christian L.

    Is 15 years of amortization realistic? I can see this for rooms, floors and power lines but not for cooling infrastructure. I don’t know any server room with even 10 year old cooling infrastructure.

  • Alessandro A.

    I agree with Christian L. 15 years of depreciation for cooling infrastructure may be fairly optimistic.

    The weight of the capital expenditure on servers may depends -a lot- on which kind of servers do you use, standard companies may use very standard and cost-effective servers like blades instead of big servers like superdome, big ibm and sun.

  • Technology has affected society and its surroundings in a number of ways. In many societies, technology has helped develop more advanced economies (including today’s global economy) and has allowed the rise of a leisure class. Many technological processes produce unwanted by-products, known as pollution, and deplete natural resources, to the detriment of the Earth and its environment. Various implementations of technology influence the values of a society and new technology often raises new ethical questions. Examples include the rise of the notion of efficiency in terms of human productivity, a term originally applied only to machines, and the challenge of traditional norms.

  • Energy storage is a very tricky topic.
    It is necessary to keep in mind that storage is the exct opposite to what energy really want tos be. Batteries, fuel cells and capacitors are a way to slow down the loss of energy, but generally the principle of energy is to act and not to be stored.

    The current technologies are unsufficient solutions to that current situation. The concept of enegerystorage needs to be rethought.

  • Just use 4-years for servers and 10-years for infrastructure and power & cooling come on top again.

  • As a building systems engineer, I have to weigh in at 15 to 20 years cooling infrastructure life expectancy. Chillers and cooling towers last 30 years, piping lasts longer, and I’ve seen plenty of 15 year-old CRACs.

  • Birddog

    James Hamilton has just left Microsoft.. Headed to Amazon…