Microsoft shows how Windows Azure can crunch some big numbers

Windows Azure is Microsoft's cloud-based server service, but Microsoft wants to show the world that it can do more than just store data for major websites. In a recent post on the Next at Microsoft blog, the company also shows how companies are using Windows Azure to crunch some big numbers in the name of medical research.

The blog post stated that this kind of work require researchers to look at a ton of data from a lot of individuals. There is a computer algorithm that can be run by PCs to help find and cut down on the number of false positives that occur in this type of work. The technique is called linear mixed models (LMMs). However, the blog points out that this kind of work can take a lot of time, as well as a lot of processing power.

Microsoft has now developed a modification of LLMs called the Factored Spectrally Transformed Linear Mixed Model (FaST-LMM). The blog states, "It allows much larger datasets to be processed and can, therefore, detect more subtle signals in the data."

Microsoft Research decided to use the FaST-LLM algorithm to check on 63,524,915,020 pairs of genetic markers that were sent to the division by the Wellcome Trust. Windows Azure handled the actual computations. The blog stated, "27,000 CPU’s were used over a period of 72 hours. 1 million tasks were consumed —the equivalent of approximately 1.9 million compute hours. If the same computation had been run on an 8-core system, it would have taken 25 years to complete."

The end result is that Windows Azure and the FaST-LLM algorithm were combined to not only go through a ton of data, but went through that information in a relatively short amount of time. More information on this kind of work can be seen in the video above.

Source: Next at Microsoft

Report a problem with article
Previous Story

Windows XP support from Microsoft ends in 500 days

Next Story

Microsoft offers help on getting Windows 8 apps published quickly


Commenting is disabled on this article.

Not only is this impressive in raw performance at the supercomputing level in a cloud setting, but it also demonstrates the Windows Azure NT technology's ability to endlessly scale as needed.

This is also a major shift in access to supercomputing power. It is now possible that rather common coding small groups and companies now can gain access to supercomputing level processing without custom code, renting out expensive supercomputers or waiting for access all at a fraction of the cost.

As Microsoft continues to extend out Azure technologies onto more server farms, access to high end computing power will become fairly cheap.

georgevella said,
How to they come up with these (cheesy?) names. Pretty impressive tho. I wonder if they are going to publish it in a paper

You can download the source code to this project and also view the paper published in nature from this link: