There are times when one looks at technology and see dramatic innovations. Some other times it is just marketing over existing features.
I am not talking about web 2.0 and ajax which really predated their names. This time I'll discuss the cloud or "elastic computing".
I will start by saying that the technology has some potential for some problems, if you need 1,000 computers for a few hours - the cloud can't be beat. Unfortunately - this is not what most problems are made of and not what most companies need. The cloud is actually not a very sophisticated virtual servers hosting service, with steep prices starting at ~$100 monthly with no persistent storage. The simple fact is that most startups can use a good hosting provider and get more bang for their buck. If a reader will try arguing scalability - I would simply say that he can get more machines at any provider, or just host his own machines and scale them as needed without facing virtualization issues. A good example today would be that one may want to use a server with SSD for a high intensity I/O machine, the cloud simply won't give him that - and will charge those I/Os at a hefty premium.
A small exception is the google app engine, which is actually a very different cloud which actually bills for CPU resources used and not simply allocated, which in turn leads to better programs. However - for the time being it is python bound and limited.
So - if I'll sum it up, the way I see it - if you are a startup and not afraid of hosting, just host and skip the cloud for now - although it would sound sexier - it is not worth it - yet.
1 comment:
And so it appears that the best clouds are still the ones in the sky :)
Post a Comment