Windows Azure Harnessed For Science

Dennis Faas's picture

Microsoft has released a new tool to help scientists and other researchers access the power of cloud computing. Project Daytona could help them analyze data in a more efficient and affordable manner.

Cloud System Uses Decentralized Computing

The project builds on Microsoft's could-based Windows Azure system. That doesn't simply mean storing information online, but also using remote computers to carry out tasks.

Cloud computing can range from relatively simple tasks, such as accessing email on services such as Hotmail, to more complex tasks, such as the way some websites run through databases that are hosted and processed by Amazon's EC2 service.

New Problems, New Solutions

The new Microsoft tool aims to solve the problem that researchers often need to do more with data than look up a specific fact. Often they'll need to hunt for patterns in the data, or carry out simulations of every possible event.

For example, environmental scientists looking at the possible effects of a river bursting its bank might need to check every possible course the excess water might take, the resistance it could meet, and ultimately the damage it could do.

Similarly, a medical researcher might need to trace the effects of a course of treatment while both ruling out any other causes of the effect, and figuring out if there are any side effects.

Immense Computer Power Costly And Unwieldy

All of this requires a huge amount of computing power: even though each individual task is straightforward, the sheer number of tasks can mean powerful computers are a necessity.

This can cause problems in two ways: firstly, such computers aren't always affordable for scientists, and if they are it can be wasteful if the computers are only needed occasionally. Secondly, using their own hardware means the scientists have to stick to one place, with can be a problem if they need to travel widely to gather data.

Project Daytona works by splitting up all the data into small chunks, spreading the chunks across multiple computers on Microsoft's Azure system that then processes the individual tasks, then recombines the data to produce the results.

What this means is that multiple machines can be working on the job at the same time, thus speeding up the process immensely without the need for a costly supercomputer. (Source: microsoft.com)

To make things simpler, Microsoft will also be offering a service known as Excel Datascope that allows users to upload data and then pick from a series of pre-programmed analysis projects. That could allow scientists to carry out research without having to figure out the computer code to achieve the analysis they want. (Source: readwriteweb.com)

Rate this article: 
No votes yet