In many documents and introductions to cloud computing we have been shown what the Cloud really means and what is possible to do inside it.  Things that have been covered have been what systems make up the cloud right up to applications and services running in it.

This blog entry aims to covers another aspect and that is computation in the Cloud.

So what does that mean?
Well going back in history a bit we started off with terminology such as super computing or clustering, to do this involved using 5 or more physical computer systems linked together over a network connection running specific software to allow all machines to run as a single super computer.  The types of software used at this time were Parallel Virtual Machine or PVM and Message Passing Interface or MPI, these were the two common standards.

No more metal systems

Today we now have the Cloud so we no longer need bare metal systems since we can run the whole thing within the Cloud environment.  Software to run clusters in the Cloud are built around a technique called Map Reduce, simply this allows a master system to accept a job requiring some work, then split that work up among his workers thus splitting up the job into smaller parts and working towards the answer collectively.

This level of computing has been proven to be very powerful from sorting huge amounts of data in minutes versus hours to being tested to predict stock exchange fluctuations. Also, the practical applications that this ideology could satisfy is huge, pretty much anything from encoding/decoding video and audio streams to computing weather predictions and generating alerts in real time.

Main benefit of cloud computing
The key main benefit of using the Cloud over the traditional methods of building super computers, is that you have the flexibility of scalability in that if the job requires more workers, then more workers can be spawned on demand to pick up the slack where needed.  This would be done on the fly very similarly if not identically to scaling a service like a web application under load.

Closing consideration
In the Cloud you only pay for what you use, there are no costs regarding infrastructure, power and footprint to worry about.  So it is expected to see large growth here with the costs making economic sense over the amount of raw computing power you could have at your fingertips.