Cloud computing is developing and becoming the place every IT company should be: either provide software through the cloud, or provide cloud facilities. And every customer is expected to use "the cloud". As usual there is some "crowd effect", and we can expect this trend to smooth down in a few years.

What is interesting here is the main argument used by cloud computing promoters: computer power is going to be distributed like power (electricity). Would you buy your own power plant for your home? We are leaving the same revolution that the one that occurred by the end of 19th century, with a power distribution network being built and progressively replacing power production installations.

Obvious isn't it?

No, it's not. Power production benefits from scale: the biggest the power plant, the higher the productivity. Because one huge turbine is most of the time more effective than many small ones, because it is not possible to build small nuclear plants, because many reasons that are well known from mechanical engineers. And so there was an obvious economical reason to centralize power production: efficiency.

The key factor at that time was that it was almost not possible to build or buy a power production installation for people, and electricity distributed through the power network was less expensive than the one home produced.

Now let's have a look to computing power. It can be easily checked that very powerful computers cost a lot; very quickly, a computer two times faster costs more than twice the price. Data center provide a lot of computing power by multiplying the number of computers, not their sizes. Lots of small steam turbines. This does not mean that there is no interest in datacenters; power, cooling, administration is more effective because of scaling benefits. And also there is an averaging of the required power, each consumer does not need to own an equipment that fits the maximum peak of power requirement.

What seems to me is that the economic benefit of cloud computing is not as high as electricity did provide a century ago. Companies cannot entirely rely on external high speed connections (as well as a hospital cannot rely entirely on distributed power). Other such revolution did fail, like heating through steam distribution that was tested at the beginning of the 20th century and did not succeed.

Anyway, there are a lot of benefits in cloud computing: standardization of hardware and systems, virtualization, green computing, many technologies developed for the cloud are generalized and benefits to all the industry. And using the cloud can prove interesting especially for small companies that cannot afford a fully trained IT team, or that have fluctuant computing power requirements. Mixing "private cloud" (I know, sounds like a joke) and real cloud is interesting as it mitigates risks and allow for a greater flexibility.

Cloud computing is a real good thing. But it is no revolution, it is not like electricity. Please stop simplifying things, marketing is a pity; give a voice to engineers.