Dead boffins deflate the cloud
History repeating itself?
Afterlife Panel Like Service Oriented Architecture (SOA) and Web 2.0 before it, the phrase "cloud computing" has become catnip to marketing zealots and lost souls of the IT industry.
Behind all the fluff, though, the cloud touches on some timeless computing topics - big systems, parallel processing, collected intelligence, and the value of expensive versus commodity hardware.
These are issues the founding fathers of IT have struggled with.
With that in mind, The Reg presents an Afterlife Panel: an imagined debate between the brains that helped make the industry what it is today.
We present the father of modern computing and advocate of intelligence Alan Turing, award-winning computer languages expert Edsger Dijkstra, and the founder of one of today's biggest vendors trying to milk the cloud IBM, Thomas J. Watson Senior.
Their topic: Is cloud computing the new mainframe gig in the sky or just personal computing reborn?
Five computers to rule them all?
Thomas Watson founded IBM - the world's largest provider of computer systems and services - but what he's really remembered for is something he never actually said about how the world will only ever need five computers.
"What I meant back in 1943 was that there was only a market for five computers at that time - a prediction which held true for maybe another decade. And, anyway, everyone makes predictions in this business that turn out to be rubbish," Watson told Turing and Dijkstra.
"I said many years ago that I have made a good many predictions about the future of our business and I have been wrong every time because I have always underestimated the possibilities."
What's got Watson's ire is the fact his famous misquote has found new life this time with the cloud. People are using him to support the theory we are on the verge of a shift away from distributed processors to centrally-controlled server farms in the form of - yes - the cloud.
"How could any one think for a moment that a bastion of capitalism like IBM would not be eager to see a demand for thousands - even millions - of computers? I didn't call the company International Business Machines for nothing. People need to 'Think' more before they speak without any real knowledge of what they are talking about," Watson stormed.
Alan Turing paused from gnawing on an apple and - in part at least - agreed with Watson. "What is really annoying about all this talk of cloud computing is that it isn't even a new idea. I gave a talk back in 1947 about a central national computer with remote terminals - even came up with a design.
"But in post-war Britain there was neither the funds nor the interest in building such a machine. Of course, your lot stole a lot of my ideas - stored programs and language compilers and so on - and went ahead and built the damn thing. Von Neuman even stole my ideas for the EDVAC."
Watson ignored the jibes - muttering under his breath about changing his company name to IBM in 1924.
"Where did you get the idea for a universal machine in 1936, then? It could be argued that you stole the name from us. But the point - as you say - is that it is not a new idea. Often the same ideas get recycled under different names or new ideas get recycled under the same names. I should know - it is the foundation of IBM's marketing philosophy."
Professor Edsger Dijkstra was torn between uncharacteristic humility for the august company he found himself in and a frustration at what he saw as the shortsightedness of Watson and Turing.
"It is sad how history repeats itself," Dijkstra opened. "In 1977, at the dawn of the so-called personal computer revolution I saw the danger that would come to pass. In the 1950s, we were told computers were expensive because they were so great. But by the 1970s, we were told microprocessors were great because they were so cheap and we began to repeat all of the mistakes on small computers that had been made earlier on large computers."
Turing interrupted: "That's all very well but I think we should concentrate on what general purpose machines can do - not what they can't do. My later work convinced me that we could use a general-purpose computer - what people have kindly named the Turing Machine - that could emulate the human brain. All this cloud stuff could enable creation of the ultimate artificial intelligence - a worldwide brain amplifier."
"Pah!" Dijkstra exclaimed. "Look Alan, your early work was brilliant and created the world of computing that we know today - but the artificial intelligence stuff was nonsense. The argument is whether it is better to distribute computing power in small computers connected by a network or centralize it in a single resource and dole it out via dumb terminals."
"We can live with either scenario - as long as we can make money," Watson said.
"Yes I am sure you can. But you - as in IBM - are totally responsible for what we have today. You designed an inadequate architecture for multi-user timesharing and inflicted it on the world in the form of the 360/370. Just because you did such a bad job doesn't mean that the idea was wrong. But then you made it worse by inflicting the PC on the world along with the opportunity to repeat the same mistakes again."
"We made an awful of money, though, and the guys who run the show today will no doubt do the same," Watson said.
By now, Dijsktra was on his high dudgeon. "The supporters of this cloud computing are seduced by the slogan that microprocessors are now so cheap that, so that when you need more processing power, you just hook a number of them together!
"The problem of how to distribute the computing load over a larger number of processors is, as far as I am aware, generality far from understood," he stormed.
Question of scale
"And even in those circumstances where the load distribution does not present a serious problem, the organization of the communication between the different processes is a problem of which only the fool talks lightly. The simplicity provided by a large central store is perhaps even more striking than the simplicity provided by a fast central processor."
Watson regarded Dijkstra paternally: "Professor, you are too much the pessimist. We have proved the worth of our machines to the world - it is an obligation for IBM to plan for its business to go on for all time.
"If that means building cloud configurations and selling them to our customers - that's what we'll do - whether it means we have five very large, very expensive computers or a million very cheap ones."
Turing had, by now, drifted off to thoughts of the implications of connecting millions of computers together, quantum interactions and the possibilities offered by the Large Hadron Collider. "The world continues to get more interesting," he said mysteriously. ®
Sponsored: Benefits from the lessons learned in HPC