Should Scientists Use Microsoft’s Free Cloud Services Offerings?

Image via Wikipedia

I am a strong proponent of using cloud computing for scientific research. Some of my posts on the topic are listed below.

• LHC Computing Grid and the New Era of Scientific Computing
• Cloud Computing and Science – Mathematica Embraces Cloud Computing
• Cloud Computing and Science – MATLAB on Clouds
• Amazon Tries to Lure Scientific Community into the Clouds
• Academic Research On Cloud Computing Gets Funded
• Scientists And Cloud Computing – Part 1
• Scientists And Cloud Computing – Part 2
• Biotech Companies And Cloud Computing

If you read these posts, you can understand how strongly I feel about the use of cloud computing in academic research, in general, and scientific research, in particular. Cloud Computing can empower scientists and help them accelerate their research while cutting down on the expenses. This is especially true in a country like US where scientific funding has been seriously curtailed since 2001. Using cloud computing is, possibly, the most efficient and cost effective way to do scientific research.

In fact, Microsoft Research has a featured story talking about how cloud computing can help scientists because of the economies of scale and an interesting comment by Prof. David Patterson, a professor of computer science at the University of California, Berkeley. According to Prof. Patterson, the potential impact of cloud computing is comparable to that of the invention of microprocessors. Absolutely fantastic comparison, in my humble opinion.

Patterson adds that the economies of scale possible with the cloud are just as much about performance as cost. The most exciting part of cloud computing, he says, is the ability to “scale up” the processing power dedicated to a task in an instant.

Even though I am happy to see Microsoft echoing some of the ideas I have been advocating in this blog for more than a year, I am deeply disturbed by one aspect of this article. It is about Microsoft’s attempt to push their S+S approach to the scientists. Here is a comment by Dan Reed, corporate vice president of Microsoft’s Technology Policy and Strategy and eXtreme Computing Group

There is a large community of researchers — social scientists, life scientists, physicists —running many computations on massive amounts of data. To use an example many people can understand — how can we enable researchers to run an Excel spreadsheet that involves billions of rows and columns, and takes thousands of hours to compute, but still give them the answer in 10 minutes, and maintain the desktop experience? Client plus cloud computing offers that kind of sweet spot.

It appears National Science Foundation has already signed an agreement with Microsoft to offer American Scientists free access to Microsoft’s cloud computing services.

The National Science Foundation and the Microsoft Corporationhave agreed to offer American scientific researchers free access to the company’s new cloud computing service. A goal of the three-year project is to give scientists the computing power to cope with exploding amounts of research data. It uses Microsoft’s Windows Azure computing system.

Even though I am fully convinced about the impact of cloud computing on Scientific research, I don’t see their S+S strategy serving the interests of Science. Well, actually, I am against this agreement for two reasons.

• In my opinion, Science has to be completely open. Any attempts to lock-in the scientific results in proprietary platforms or applications or data formats goes against the very spirit of openness in science. By locking in the scientific results within Microsoft’s platform, we are forcing the entire scientific community to be dependent on Microsoft, affecting the very advancement of science itself. This is morally wrong and very bad for the scientific community in the long term.
• Secondly, S+S approach is promoted by Microsoft to protect their cash cow. It adds problems in two ways for the scientific community. First, It is not cost effective and, more importantly, it is not an efficient way to do science either. The exorbitant licensing costs for the traditional software used to access these cloud services will put a huge dent on the funding for scientific projects. In my opinion, the huge amount of money spent on such traditional software packages can be better utilized elsewhere. Then, there is the issue of a need for bigger resources to use these bulky software. These resource needs not only make this S+S approach inefficient, it also adds to the cost of computing. What is the point in saving money on the traditional infrastructure expenditure and then spend a part of it on traditional extra powerful desktop (laptop) computers?

Another important point to note from the NYT story is that this agreement between NSF and Microsoft is for three years. In the absence of an indefinite free use agreement, this is just a pure marketing ploy. After three years, these research projects will be forced to spend quite a bit of money on the licensing fees for the traditional software plus the cloud services offered by Microsoft. Not only that, any researcher who wants to either extend these projects or use their results on new projects may end up paying part of their funding money to Microsoft.

In short, this agreement between NSF and Microsoft is very shortsighted and it may as well go against the very spirit of science. I have no problem with Microsoft pushing their S+S strategy in the market. Eventually, the economics and the value add will determine whether S+S or pure SaaS will be the ultimate winner. However, when it comes to altruistic issues like Science where public money is heavily involved and its impact on the public is very significant, it is not a good idea to push a strategy that serves a particular company’s self interest alone. This move is neither good for Science nor for Cloud Computing.