Global Warming

12th June
written by simplelight

One of the major issue with large data centers is power. This is applicable to both large data centers like Microsoft / Google and also to large Enterprise Data Centers which are very energy inefficient.

Definition of Power Effectiveness: Data Center Power Usage Effectiveness (PUE) is defined as the ration of data center power to IT (server) power draw. Thus a PUE of 2.0 means that the data center must draw 2 Watts for every 1 Watt of power consumed by IT (server) equipment. The ideal number would be 1.0, which means there is zero overhead. The overhead power is used by lightning, power delivery, UPS, chillers, fans, air-conditioning etc. Google claims to have achieved a PUE of 1.3 to 1.7. Microsoft runs somewhere close to 1.8. Most of Corporate America runs between 2.0 and 2.5.

A typical large data center these days costs in the range of $150 Million to $300 Million depending upon the size and location. A 15 MW data center facility is approximately $200 million. This is the capital cost so it is depreciated over time.

Most of the facility cost is power related. Anywhere  from 75% to 80% of the cost is power (pdu, chiller, ups, etc).

A typical 15MW datacenter with 50,000 servers costs about  $6.0 million per month for operating expense (excluding people cost) and the share of power infrastructure (pdu, chiller, ups, etc) is between 20% to 24% and actual power for the servers is 18% to 20%. Thus total power cost is between 38% to 44%. These numbers reflect what Microsoft / Google would do. EPA has done a study and they believe these numbers are close to 50% for inefficient data centers.

21st April
written by simplelight

I frequently meet with the executives of alternative energy companies. Unfortunately, in most cases, the price of the electricity they’re generating is too high or the payback period for the required capital investment is too long.

In the past two years (April 2007 to March 2009) my wife and I have used 13,226 kwh at an average price of 13.3 cents per kwh (including all the taxes, distribution costs etc.). The total cost was $1,761 which amounts to $2.42 per day. We are fairly efficient and only use about 18.2 kwh per day which, according to PG&E, is a lot lower than average.

Nevertheless, it is hard to contemplate spending $10-20k to install any type of edge electricity generation when we can so easily pull it off the grid.

California, though, has a tiered rate system. Similar to tax rates, the more electricity you consume the higher your rate per kwh. The following are the approximate rates in California as of today.

  • Baseline
  • Tier 2 (up to 130% of baseline) — 13 cents per kwh
  • Tier 3 (up to 200% of baseline) — 21 cents per kwh
  • Tier 4 (up to 300% of baseline) — 37 cents per kwh
  • Tier 5 (over 300% of baseline) — 44 cents per kwh

The baseline rate varies by geography and season but is priced around 11.5 cents per kwh and includes about 23 kwh per day currently.

If you are consuming enough electricity to be hitting tiers 4 and 5 then it begins to make sense to install solar panels or fuel cells such as those provided by ClearEdge Power.

24th June
written by simplelight

I was at a conference this morning where Spansion and Virident were presenting their latest flash memory technology designed to replace DRAM in web servers. Some interesting facts:

  1. Cooling and power distribution losses account for 50% of the electricity consumed in US datacenters.
  2. Datacenter power use doubled from 2000 to 2005 and will almost double again by 2010. Growth in electricity use has been slowed somewhat by the advent of server virtualization over the last few years.
  3. US datacenters use more electricity than countries like Sweden and Iran
  4. Datacenters use almost 100 billion kilowatt hours each year at approximately $0.10 per kilowatt hour. Datacenter electricity consumption is growing at 15% per year (!)
  5. Datacenter memory (DRAM) uses 2x more electricity than the total capacity of US solar panel installations.
  6. US, EU and Japan use 3/4 of the world’s electricity.

It will be interesting to see whether Spansion’s newly announced EcoRAM can put a dent in these problems. They are citing some impressive numbers:

  1. 1/5th the power of DRAM at comparable read performance.
  2. 800x faster than NAND flash access times.
  3. 30 mins to write 1TB of data on EcoRAM vs 5 hours using traditional NOR DIMM’s.

On the other hand, the representatives from Intel and AMD certainly weren’t giving their unqualified support to EcoRAM.

1st May
written by simplelight

I’ve heard before that walking a mile emits more carbon than driving a mile so decided to investigate for myself.

A 190lb human walking at 3 miles per hour would expend 302 kCal. Therefore, covering a mile would take 20 mins and approximately 100 kCal. Apparently it takes 7-10 kCal of fossil fuel to produce 1 kCal of food. The upper end of the range assumes a diet of highly processed food and meat being transported long distances. The lower end of the range assumes a more vegetarian-centric diet and locally produced food.

Therefore,  a 1 mile walk requires 700 – 1,000 kCal of fossil fuel energy.

Driving a 30 mile per gallon car for one mile at 60 mph takes 1 minute and 1/30th of a gallon of gas. One gallon of gasoline contains approximately 30,000 kCal of fossil fuel energy.

Therefore, a 1 mile drive requires approximately 1,000 kCal of fossil fuel energy.

Depending on where you source your food and what you eat, it doesn’t make much difference in terms of carbon emissions. Driving a car one mile is far cheaper than walking a mile. But the health benefits of walking a mile far outweigh the cost.

21st April
written by simplelight

The Wall Street Journal had an article a few days ago about the unreliability of some of the numbers behind global warming. The article goes on to say:

The fear of a sudden loss of ice from Greenland also makes a lot of news. A year ago, radio and television were ablaze with the discovery of “Warming Island,” a piece of land thought to be part of Greenland. But when the ice receded in the last few years, it turned out that there was open water. Hence Warming Island, which some said hadn’t been uncovered for thousands of years. CNN, ABC and the BBC made field trips to the island.

But every climatologist must know that Greenland’s last decade was no warmer than several decades in the early and mid-20th century. In fact, the period from 1970-1995 was the coldest one since the late 19th century, meaning that Greenland’s ice anomalously expanded right about the time climate change scientists decided to look at it.

Warming Island has a very distinctive shape, and it lies off of Carlsbad Fjord, in eastern Greenland. My colleague Chip Knappenberger found an inconvenient book, “Arctic Riviera,” published in 1957 (near the end of the previous warm period) by aerial photographer Ernst Hofer. Hofer did reconnaissance for expeditions and was surprised by how pleasant the summers had become. There’s a map in his book: It shows Warming Island.

The mechanism for the Greenland disaster is that summer warming creates rivers, called moulins, that descend into the ice cap, lubricating a rapid collapse and raising sea levels by 20 feet in the next 90 years. In Al Gore’s book, “An Inconvenient Truth,” there’s a wonderful picture of a moulin on page 193, with the text stating “These photographs from Greenland illustrate some of the dramatic changes now happening on the ice there.”

Really? There’s a photograph in the journal “Arctic,” published in 1953 by R.H. Katz, captioned “River disappearing in 40-foot deep gorge,” on Greenland’s Adolf Hoels Glacier. It’s all there in the open literature, but apparently that’s too inconvenient to bring up. Greenland didn’t shed its ice then. There was no acceleration of the rise in sea level.

Finally, no one seems to want to discuss that for millennia after the end of the last ice age, the Eurasian arctic was several degrees warmer in summer (when ice melts) than it is now. We know this because trees are buried in areas that are now too cold to support them. Back then, the forest extended all the way to the Arctic Ocean, which is now completely surrounded by tundra. If it was warmer for such a long period, why didn’t Greenland shed its ice?

This prompts the ultimate question: Why is the news on global warming always bad? Perhaps because there’s little incentive to look at things the other way. If you do, you’re liable to be pilloried by your colleagues. If global warming isn’t such a threat, who needs all that funding? Who needs the army of policy wonks crawling around the world with bold plans to stop climate change?

It seems to me that we should have thought about this before we starting using our corn to power our cars. Because that’s the kind of stupidy that leads to food riots around the world.