Cloud computing seems like the perfect research tool
Cloud computing seems like the perfect research tool: Ubiquitous access, elastic capacity, cross-platform support and it’s cheap. How cheap? Well, for National Science Foundation (NSF) researchers it can be free. Last week Microsoft announced $5M in grant money for Azure use to worthy research projects. Not to be upstaged by Microsoft, Google announced its own focused research grants with a subsidy of $5.7M for research. This sounds like a lot of money but what does $5M really buy in today’s cloud computing dollars? The vision for NSF and Microsoft is to offer a dynamic cloud computing environment for individual or groups of researchers. From the Windows Azure blog, the intent is to provide a “massive, highly efficient cloud infrastructures, such as those possible with Windows Azure Platform [that] can transform how research is conducted and accelerate scientific exploration…” So, let’s talk about something massive like processing data from the Large Hadron Collider (LHC). How much compute power does one need? Well, the Open Science Grid Consortium (OSGC) offers the equivalent of 300,000 computing hours per day for LHC analysis to groups of researchers. This equates to $36,000 in Azure dollars ($0.12/compute hour) per day. If we disregard storage and communications fees $5M buys 138 days of OSGC-equivalent processing time. Assume that Microsoft is offering Azure at cost to NSF grantees then we’re probably looking at about 275 days, or more than a year of research computing. In the end, $5M sounds like a pretty good bang for the buck.
