When there is complex processing requirement and at the same time with a requirement of high response throughput, Grid computing could be a solution on your plate. Grid computing offers powerful computers interconnected and share processing of a request. The request is parsed into chunks and distributed to the processors. The result from the processors are aggregated and form the response. Grid computing requires a facility to host these computers in a restricted environment to manage and configure. It is a costly affair and is generally used where information processing is critical and involves complex computations.

But internet has made grid computing easy in certain ways. For example Seti@home is a project that collects information from satellites and analyzes them to search for extraterrestrial intelligence. It uses the idle processor time of the personal PC’s spread across the world connected to the internet. Once the user agress to the terms and conditions, an agent is downloaded and installed on the personal computer. This agent typically runs as a screen saver and it uses the computer’s processor when it is idle time (rememer the coffee breaks and lunch time?). Once a unit of work is complete it sends it back to the server for aggregation with other completed units and receives another unit of work for processing. This type of computing is called scavenging grid. The advantage is that it is the cheapest form of grid computing but the scope where it can be applied is limited and is also dependent on the internet user’s acceptance to run it. And another disadvantage is the time it takes to process a complete request. World Community Grid is yet another example of scavenging grid. It aims at processing complex scientific research to determine cure for diseases like aids.

Yet another common grid computing type is the data grid computing. Data grid computing is used when volumnious data needs to be processed. The data grid computing is supported by high end database systems that can store volumnous data and distribute the data to different processors for processing. For much detailed explanations on grid computing refer the wikipedia.

It is usually a common debate on the topic of describing software architecture. The common questions that come up are,

1. What is architecture?
2. What is the software architecting process?
3. What is the scope of architecture in the overall development of a project?
4. When does it start and where does it end?
5. Is architecture a science or an art?

Here is an article in IBM developerWorks by Peter Eeles on the process of software architecting. And here is the follow-up summary of the same article in Artima by Frank Sommers.