Small cloud vs big clouds

I’ve been thinking about a particular future business and its infrastructure for almost 3 years now. Cloud technologies are quite relevant to what I have in mind, and recently I’ve started to think about working on a set of open source cloud implementaitons.

However, there is a problem. It is not a technical problem, it is an economic one. Amazon, Google and probably MS will be in cloud business in a quite strong way in a couple of years. Amazon and Google do it already, and the cost efficiency of hiring their infrastructure vs creating mine is very, very relevant. Cloud loves hardware, in fact its advantage is at joining rather simpler models of storage and processing with very efficient scaling. So the advantage is about scalibility, and that is dependent on hardware. Now, knowing how to design and build a particular solution to a processing intentsive problem is a valuable asset, but what if your customers do not have the money to buy the hardware that can give the performance you’d like to provide? In these kind of situations hiring cloud capacity from these giants and putting your know how on top of if is more efficient in terms of cost, and many cases will make your services more affordable.

I can see a rough segmentation of market, where in some segments clients prefer rented cloud infrastructure simply for cost benefits, and in others they choose to buy their own farm, either because their data is very sensitive or simply because they have the money.

The other problem is, serious open source cloud technology comes from the giants, like Google who has given us the bigtable, and Yahoo who has improved and tested Hadoop. These companies have the ability to develop these kind of solutions since they need it, and they have to infrastructure and use cases to test it. Without these real life connections, how can a disconnected open source initiative develop alternatives to commercial offerings?

In short, cloud belongs to big names for the moment, and it is alive as open source because they want to keep it alive.  What if this changes in the future? I guess anyone in the field will have to know about these services quite well, just to make sure they can offer it as a lower cost alternative, at least for some cases.