Serving Space
Tom Vanderbilt – author of the excellent book Survival City: Adventures Among the Ruins of Atomic America, as well as the recent Traffic, and subject of a short but interesting interview in The BLDGBLOG Book – has a long article out in the The New York Times Magazine about the architecture (and energy implications) of large-scale data centers.
This is the world of "increasingly large, powerful, energy-intensive, always-on and essentially out-of-sight data centers" that now dot the global landscape.
What is this new type of space? "Call it the architecture of search," Vanderbilt writes: "the tens of thousands of square feet of machinery, humming away 24/7, 365 days a year – often built on, say, a former bean field – that lie behind your Internet queries."
Such buildings often blend in with the everyday urban landscape. For instance, Vanderbilt describes "NJ2, a data center located in Weehawken, N.J., just through the Lincoln Tunnel from Manhattan." It is "an unmarked beige complex with smoked windows"; inside it "hum the trading engines of several large financial exchanges."
The interesting thing here is that the machines stored inside NJ2 are stored there so that they can be as close as possible, geographically, with other machines: the ones that handle trades on Wall Street. Spatial proximity, in this case, cuts down on information-relay time, thus enabling large-scale financial processes to unfold nearly in real-time.
We might say, then, that the built environment you see here – the distances between buildings and their urban or geographical locations – is thus an articulation not of architectural theory or of the stylistic assertions of one particular architect, but of the processing power of today's supercomputers.
Future changes in processing speed might then ramify outward to further tweak the built environment.
Vanderbilt explains that when the Philadelphia Stock Exchange moved its computers north, into NJ2 – a distance, we read, of 80 miles – they saved three milliseconds on every trade. Lest we laugh that off as the spatial equivalent of obsessive-compulsive disorder, we're told that "it is estimated that a 100-millisecond delay reduces Amazon’s sales by 1 percent."
It's an awesome article – check it out if you get a chance.
This is the world of "increasingly large, powerful, energy-intensive, always-on and essentially out-of-sight data centers" that now dot the global landscape.
What is this new type of space? "Call it the architecture of search," Vanderbilt writes: "the tens of thousands of square feet of machinery, humming away 24/7, 365 days a year – often built on, say, a former bean field – that lie behind your Internet queries."
Such buildings often blend in with the everyday urban landscape. For instance, Vanderbilt describes "NJ2, a data center located in Weehawken, N.J., just through the Lincoln Tunnel from Manhattan." It is "an unmarked beige complex with smoked windows"; inside it "hum the trading engines of several large financial exchanges."
The interesting thing here is that the machines stored inside NJ2 are stored there so that they can be as close as possible, geographically, with other machines: the ones that handle trades on Wall Street. Spatial proximity, in this case, cuts down on information-relay time, thus enabling large-scale financial processes to unfold nearly in real-time.
We might say, then, that the built environment you see here – the distances between buildings and their urban or geographical locations – is thus an articulation not of architectural theory or of the stylistic assertions of one particular architect, but of the processing power of today's supercomputers.
Future changes in processing speed might then ramify outward to further tweak the built environment.
Vanderbilt explains that when the Philadelphia Stock Exchange moved its computers north, into NJ2 – a distance, we read, of 80 miles – they saved three milliseconds on every trade. Lest we laugh that off as the spatial equivalent of obsessive-compulsive disorder, we're told that "it is estimated that a 100-millisecond delay reduces Amazon’s sales by 1 percent."
It's an awesome article – check it out if you get a chance.
Comments are moderated.
If it's not spam, it will appear here shortly!
It's a shame that Tom V. has fallen into the now-common 24/7/365 trap. It should really be "24/7, 52 weeks a year", not "24/7, 365 days a year", shouldn't it? Otherwise what's the '7' in '24/7' doing there?
drumbux - Ah, but working 7 days a week for 52 weeks would only be 364days a year!
I'd be interested in a map of mega-scale data centres in the US and, perhaps moreso, worldwide. I wonder what oddities might turn up in their distribution?
Reminds me of Grace Hopper giving David Letterman a nanosecond.
Post a Comment