Microsoft, the Times reports, is offering scientists free access to its cloud computing. This is important because scientists are grappling with mountainous troves of data, and they need Google-like (or Bing-like) computing clusters to crunch them. I read recently that the biological data amassed last year surpassed all of the biological data in history, presumably from the dissection of the first frog until weeks before the Obama inauguration.

The need for scientific clouds is clear, and as I wrote two years ago in BusinessWeek, IBM and Google are on a similar track. My question is this: Is scientific data going to get tangled in a software battle? IBM, Google and others are offering an open-source cloud software known as Hadoop, which is based on Google's MapReduce. Microsoft is providing its own platform, Azure. The grand promise of cloud computing will be for scientists to share data sets, and even to delve into ones from seemingly unrelated fields. That way they might find correlations between, say, meterology and disease.

But if scientists in the Microsoft cloud are doing their work in Azure, will they be able to collaborate with others working in the cloud? The last thing science needs is a platform battle in the next generation of computing. (This thread of questions on a Microsoft site shows developers grappling with the challenges of offering Mapreduce within Azure.)

I also notice that the Microsoft-NSA grant is for U.S. scientists. I'm assuming researchers from elsewhere will have access too. Researching teams in science stopped paying attention to borders long ago.


Wondering what scientific cloud computing looks like? Rob Gillen, a researcher at Oak Ridge National Laboratories, explores some meteorology data using Azure and other Microsoft technologies, including the Surface touch screen. (This version is more for the presentation of cloud data, I would assume, than the nuts-and-bolts of actual research.)

(embeded object)



Link to original post