The Information Factories Are Here

prostoalex writes, "Wired magazine has coined a new term for the massive data centers built in the Pacific Northwest by Google, Microsoft, and Yahoo! Cloudware is, ironically, a return of the centralized data and bandwidth power houses caused by the decentralized and distributed nature of the Internet. George Gilder thinks we're witnessing something monumental: 'According to Bell's law, every decade a new class of computer emerges from a hundredfold drop in the price of processing power. As we approach a billionth of a cent per byte of storage, and pennies per gigabit per second of bandwidth, what kind of machine labors to be born? How will we feed it? How will it be tamed? And how soon will it, in its inevitable turn, become a dinosaur?'"

Comments

Supply of fiber too low for a revolution?

[Salvance]

At some point massive data centers won't provide incremental benefits unless the massive increases in processing power are met with proportional decreases in bandwidth prices. Sure, bandwidth prices have dropped, but not by nearly the rate of price/teraflop processing has. Companies like Google recognize this, and are investing in their own fiber to compensate. But the telecommuncations companies are the ones that originally build these lines, and it's unfortunately in their best interest to keep the supply of spare bandwidth very low.

Why Gilder Is Telecosmically Wrong

[miller60]

Everything is getting cheaper but power, which for some data centers now costs more than hardware. Nicholas Carr explains why Gilder's assumptions are problematic:

"What Gilder calls 'petascale computing' is anything but free. The marginal cost of supplying a dose of processing power or a chunk of storage may be infinitesimal, but the fixed costs of petascale computing are very, very high. Led by web-computing giants like Google, Microsoft, Amazon, and Ask.com, companies are dumping billions of dollars of capital into constructing utility-class computing centers. And keeping those centers running requires, as Gilder himself notes, the "awesome consumption" of electricity"

As I noted in our commentary at Data Center Knowledge, the power issues with high-density blade server computing has been understood for years. Back in 2002, Liebert and APC and other equipment vendors were developing products that could address huge heat loads. They saw it coming, and sensed a market opportunity. So where were the chip makers? Even as cooling vendors prepared for the results of the huge power and heat loads, little was done to address their source.

It doesn't matter

[sillybilly]

It doesn't matter if computing performance doubles if the software that runs on it decays in performance at an even greater rate. Back in 1988 MSDOS used to boot in less than 10 seconds after the BIOS POST. Who cares if you'll have software with features greater than your brain, with capacity to even guess your thoughts, wishes and desires, if it will just do what you want without mouseclicks of speech commands, who cares for all these features if it takes 5 years to boot up on a computer a gazillion times faster than today's computers, and its processing speed is uttering 3 words per decade while consuming 900 gigawatt of electric power? Case in point: Windows Vista.

I'm sorry

[Colin Smith]

The future will eliminate that differentiation. Data will not be 'here' or 'there'. Rather, it will be. Data will simply exist and we will access it as if it were immediately 'here' all the time.

But this is the biggest load of new age bullshit I've heard in years.