RAD Lab, “Above the Clouds: A Berkeley View of Cloud Computing,” UC Berkeley Technical Report UCB/EECS 2009-28, 2009. [PDF]
Summary
The term ‘Cloud Computing’ has been in extensive use for last few years to suit many different, often unrelated, needs. This report tries to define what Cloud Computing might mean by identifying its characteristics in terms of hardware and software, quantifying economic tradeoffs, and analyzing the top ten obstacles (toward its adoption and growth as well as its policy and business implications) and corresponding opportunities to overcome those obstacles.
The authors define Cloud Computing as a more general instantiation of Software as a Service (SaaS) that can provide Utility Computing to third parties that do not own their own datacenters. They refer to the datacenter hardware and software as a Cloud and differentiate between Public Cloud (available for other’s to rent) and Private Cloud (belongs to individual entities such as large Internet service providers). The authors also identify three hardware aspects that are completely new in Cloud Computing: the illusion of infinite computing resources that can be dynamically scaled up or down on the fly.
This report compares Cloud Computing vendors across three different axes: computation, storage, and networking, but keeps their economic models separate. Each of the existing Cloud providers (e.g., Amazon EC2, Microsoft Azure, and Google AppEngine) take different routes in the ways they expose these resources to their customers. It is very hard to say if one is better than another, and the authors try not to make such an statement; however, there are implicit nods toward a model that has the least amount of restrictions (read Amazon EC2). I personally agree with this because for my (systems) research I want to be able to have as much control over the resources I am using as possible.
The most interesting part of the report is the list of obstacles and corresponding research opportunities (Table 6). It has been only more than two years this report has been published, but several of the opportunities mentioned here have already been given rise to numerous research papers. The >1100 citations (according to Google Scholar) this technical report has gathered is a testament to its appropriate timing and large appeal, which will continue to be steady in coming years.
Critique
This is not a technical work, rather a comprehensive study that tries to sort out a workable definition of Cloud Computing. Instead of trying to impose yet another definition, the authors dig deeper to understand the key characteristics of the Cloud and put forth the tradeoffs and opportunities in front of the reader for them to decide. This is a must-read for all techies and many non-techies out there.