Slashdot Mirror
The Uncertain Promise of Utility Computing
icke writes "A quick overview of where the Economist thinks we are with the The Next Big Thing, also known as Stuff that doesn't work yet. Quoting: 'It is increasingly painful to watch Carly Fiorina, the boss of Hewlett-Packard (HP), as she tries to explain to yet another conference audience what her new grand vision of "adaptive" information technology is about. It has something to do with "Darwinian reference architectures", she suggests, and also with "modularising" and "integrating", as well as with lots of "enabling" and "processes". IBM, HP's arch rival, is trying even harder, with a marketing splurge for what it calls "on-demand computing". Microsoft's Bill Gates talks of "seamless computing". Other vendors prefer "ubiquitous", "autonomous" or "utility" computing. Forrester Research, a consultancy, likes "organic". Gartner, a rival, opts for "real-time". Clearly, something monumental must be going on in the world of computing for these technology titans simultaneously to discover something that is so profound and yet so hard to name.'"
She has enough money in her coffers (thanks to over 6000 layoffs translating to a $150M bonus last year) to give everyone she's ever met the finger, buy an island somewhere near the equator, and sip margaritas all day every day until she dies a miserable and lonely death.
She knows nothing about technology, and rather little about business. She only knows how to drain money. Don't expect to see HP change the face of computing with her in the captain's chair.
Cretin - a powerful and flexible CD reencoder
The instructor couldn't explain it, so she brought in a marketing exec, who could only define it in terms of itself. "E-business on demand is about computing, on demand, for e-business." Sprinkle in a healthy dose of meaningless adjectives, and you get the picture.
I'll tell you, it's pervasive. Since then, I've not found one person who can give a cohesive definition at this company. And yet, it's supposed to be my driving force and ultimate goal.
yay.
I was at Thinking Machines (the company that invented massively parallel computing) a decade ago, and back then Danny Hillis talked frequently about "utility computing" -- the idea that your computations would know how to flow back to wherever it needed to be done. So you'd work on a desktop computer and the user interactive bits would run locally, harder parts would flow back to a big CPU in the basement, and the really hard parts could flow back to a city supercomputer, in a CPU equivalent of the power grid.
...
At a high level, it's a pretty simple idea, and very powerful.
At the detailed level, there are some amazingly hard problems to solve. Like, for example, how does software get split into parts that can be separated with minimal communications overhead, or how do you decide when a task would run more efficiently spread across a bunch of CPU's, or how do you keep running smoothly when a network outage causes 10% of your CPU's to drop off of the grid.
I suspect that the reason that all of the big companies are pitching this is that:
1) CPU's and operating systems have been commoditized by Intel/AMD/etc. and Linux, and they want to have a reason for you to buy bigger/better/more expensive systems.
2) Once one of them announced it, they all have to have a "response".
That being said, I think that what they're doing is going to be of real value to high-end customers. If you're running a farm of 5,000 servers, you really need the software to be self-healing, etc.
Enable 3D printed prosthetics!
Nope, sorry. On-demand means that the computers can scale to cope with any load, in real-time.
Companies will not own their hardware, but rent it. If they suddenly need 3 times as much CPU, then they get it immediately, and only pay for what they use.
This is different than the current situation where a company must always keep enough hardware around to handle peak loads, which is almost never. And then, if they guessed wrong, they are still screwed.
It's really that simple, but hard to implement. IBM plans vast server farms, and large-scale software migration projects to handle cases where a customer uses 37 different databases, include 14 distinct versions of Oracle. Port all the databases to one version of Oracle or DB2, and suddenly you can scale your database capacity much more easily, and in real-time.
Well, I think they're flogging a bit more than just that: they're envisioning more widespread use of distributed computing. Distributed computing, according to these market leaders, will enable companies to come up with a working, marketable, and profitable way to sell computing power to other companies through "utility"-like means (think "metered", like electricity).
As for what that means for us *normal* people, maybe it means we can opt to make an extra penny or two running an IBM branded screensaver that runs computations for them (kind of like SETI@Home, just profitable). Or maybe it means we'll be forced to do the same for free if we use a Microsoft product =). Who knows?
I think that a lot of people are missing that there are two concepts here. The first is grid computing, which is as far as I understand being able to offload processing to multiple computers. The second is ubiquitous computing, which is being able to use computers anywhere you want and access data anywhere you want in a natural fashion such that you're not even thinking about the fact that you're using a computer. See this google-cached page for an example. The two may be used together but are not dependent on each other.