Nikita Ivanov writes about grid computing, which you normally associate with big iron. However, Nikita has written about grid computing with Android, and how, with power mobile phones, the largest grid out there could be the cell phone market:

I heard this idea probably 3-4 years ago when somebody mentioned to me that the biggest grid in existence is the grid comprised of all the cell phones out there. Think about it: every cell phone acts as a grid node, it has certain (albeit minimal) processing power and can easily communicate with any other cell phone in the world peer-to-peer (well, in practical sense, almost easy). And there is a special class of grid tasks that fits this type of grid perfectly: hyper-parallel grid tasks - the tasks that can easily split into thousands or even tens of thousand small micro-jobs where each such micro-job can take 2-3 seconds to get delivered onto processing device and spend another 5-10 seconds on it to get processed and send optional result back.

Now, back 3-4 years ago there were several obvious problems in this picture:

• Phones were hopelessly underpowered and worked performed by a grid of hundreds of cell phones could easily be done by a single workstation
  • Not so today. Phones are quickly getting more and more powerful and there is a class of tasks that can fit naturally into this profile (like human image recognition, for example).
• Connectivity was so slow that it basically negated the whole parallelization effect as grid would have been spending all its time in communication
  • Not so today. 3G phones are (finally) getting to US and within 2-3 year will dominate the phone market. With 3G connectivity – the problem is basically solved as you are getting your office type of connectivity on your cell phone.
• Phones had proprietary system APIs and development of the uniform grid middleware was very hard if not impossible all together.
  • Not so today – and that’s the key.

Can you feel your battery sucking away as it tries to find aliens?