As I’ve been watching the evolution of the cloud, it has become clearer and clearer that a huge transition is in process, the transition from isolated computers that send data to one another over networks to a highly integrated worldwide computing fabric that distributes computation, storage, and data movement seamlessly, dynamically, and automatically without (much) regard to geography. This endpoint is what I've called for some years the "world computer.”
I'm not the first or only person to think about this. I remember my Dad told a joke when I was a kid about the scientists putting together the first galactic computer. They ask it a question: "Is there a God?” The computer replies: "There is now.”
However you view it, it's hard to think intelligently about the world computer, as a technologist, as an investor, as a policy-maker, or as a citizen.
One big set of questions has to do with how -- and how fast -- the evolution will occur. Investors like me care about these questions a lot, because we need to make profitable bets on the right pathways to cloud computing at the right time. If you invested in magnetic bubble memory in 1980 (which was "sure" to be the persistent storage solution instead of those flakey, slow floppy disks) or if you invested in SNA in 1992 (because it was the only "real" networking standard, unlike that sloppy bastard child TCP/IP), you were out of luck. If you bet on Netscape in 1996 or MP3 players in 2004 you were equally out of luck, because it was the wrong time.
Most people would agree that we will reach the world computer stage somewhere in the not-too-distant future, but there are a lot of questions about 1) how long it will take, 2) by what routes it will develop, and 3) how to invest profitably in this truly big wind.
Here are some of my thoughts on these questions:
- Things will happen quickly. I think that the general course of “cloud-i-zation” is going to go faster than others seem to think. I see us being essentially at a world computer status in five to ten years.
- Geography will matter. There will be regional clouds based on minimum latencies and possibly some kinds of regional specialization. So, we might have an East Coast North American cloud with some specializations in financial and security ops. Maybe this cloud is based in Iceland and services London as well? There might be a "supply-chain" cloud based in Taiwan or China. There might be an "entertainment/media" cloud based in LA.
- We are "crossing the chasm" into cloud-ville today. Early adopters are trying to get ahead of the curve for advantage, but early majority customers do not see a credible "complete solution" yet. Geoffrey Moore, the author of Crossing the Chasm, teaches us that complete solutions arise in some niche or beachhead. The profitable investments today will be cloud-oriented solutions for that niche or niches.
- Cloud appliances will emerge. One possible kind of niche for a complete solution, I think, is what I call the "cloud appliance." Like the hardware appliance for which it is named, a cloud appliance is a cloud-based solution that performs one kind of optimized function well.
- Ad exchanges will take the form of cloud appliances. An ad exchange is a good example of a cloud appliance. The utility of having a cloud-based ad exchange is pretty clear, provided that the bidders are located "latency-close" to the exchange.
- Email and email archiving seem like two other natural cloud appliances. There is a widespread migration of email from premises-based to cloud-based today, I think; you see signs of it all around. Maybe too late for the email appliance slot (which may be occupied by Google (Nasdaq: GOOG) and Microsoft Corp. (Nasdaq: MSFT), among others), but perhaps the email archiving slot is still open. An email-archiving cloud appliance would manage both premises- and cloud-based email, would archive to cloud storage, and would manage e-discovery of the email archive from a Web front-end.
There are equally vexing questions about the world computer for policy-makers and citizens. Grist for another blog.
— Dan Gordon is director of research at Valhalla Partners. He has over 30 years experience working with technology, as a computer scientist, software developer, manager, analyst, and entrepreneur.