The cloud will be huge at some point, right now the numbers simply aren't there and there have been too many high profile glitches in recent times to get fence sitters to switch in large numbers.

Comparing the costs of cloud hosted computation and storage facilities with a 100 Mbit flat rate box with 500G on board for $130 / month is a good exercise and seems to indicate that cloud storage and bandwidth are much too expensive to be competitive except for some very specific uses.

Price is of course only one factor in making this decision (reliability, scalability are the others), but as someone put it nicely building your small web app on top of the cloud infrastructure is essentially a case of premature optimization.

It's a time of transition, maybe in the long run everything will be 'cloud based', but I think there is definitely a hybrid possible, where the same architectural tricks that go in to cloud computing are used by hosting centers to facilitate easy deployment.

Portability between suppliers is another key sticking point, right now the cloud seems to be more about locking in customers than anything else.

The only thing the cloud can do cheaper right now is massive amounts of compute power for short periods of time. If you need to do something every month which takes 1000 CPU-days, but needs to be done in a day for the results to be of use, you've either got to buy lots of hardware or use the cloud.

The ability to do this cost effectively could make certain computations viable, opening up new opportunities for people to do interesting and useful things.

Yes, exactly, that is what I had in mind with my 'specific uses'.

"building your small web app on top of the cloud infrastructure is essentially a case of premature optimization."

Ah, I thought it was the opposite... at first you use the cloud to take advantage of the infrastructure at a low cost and with much less management and then when you get big you start hosting your own stuff because it's cheaper and it's worth it. (I'm ignorant on those matters)

I don't think the question is whether it can be trusted or not. That's pretty ambiguous and, as mentioned, you can't really put 100% trust in anything. Also, the meaning of 'the cloud' has become diluted, seeing it's now analogous with pretty much anything web based. We might as well call it "scalable, distributed and virtualized computing" or something.

The question, for me, comes down to education and understanding. The term 'the cloud' is relatively nascent, but the basic underlying principles are not really that new, just more evolved. The cloud is as secure and reliable as you make it. By definition it promotes redundancy, accessibility and scalability.

I work on a 'cloud' claims management application and it's targeted at the enterprise (finance/insurance) market, who are notoriously slow to adopt new technology. Trust me, it's a tough audience, but not as tough as you might think. The realization that data can be accessed (and work done) anywhere, at negligible cost with existing hardware is simply too compelling for any forward thinking company.

Quite simply, the pros seriously outweigh the cons - whereas the opposite is true for traditional client-server technology.

I don't think client-server is the alternative to the cloud. The only company-wide client-server application I saw in most organizations was Outlook. The rest, from time tracking to individual performance evaluation, was all done in the intranet. So the real choice is between hosting in the cloud or self-hosting.

Just like the goldbugs say (in respect to owning physical gold instead of holding it in ETFs), "If you don't hold it, you don't own it."

But can anything be entirely trusted?