Scaling and benchmarking application frameworks in the cloud. How will popular development environments like RoR scale/perform on modern (manycore) platforms and on “cloud computing” platforms such as Amazon EC2? We are collaborating with RAD Lab founding sponsor Sun Microsystems on investigating scalability and performance of RoR and other Web platforms, on both “cloud computing” environments and modern multicore hardware. We have made many improvements to Sun’s Olio, now an Apache Foundation Incubator project, have suggested a metric and methodology for evaluating its performance in different deployment environments, and are using it internally as a representative Web 2.0 test app (it includes its own workload generator) for a number of projects.
- Undergraduate Students: Aaron Beitch, Timothy Yung, David Mills (visiting SUPERB student), Arthur Klepchukov (graduated 2009), Hubert Wong (graduated 2009, now at Amazon.com), Jimmy Nguyen (graduated 2009)
- Collaborators: Will Sobel (visiting lecturer until May 2009, now at SystemInsights.com); Shanti Subramanyam, Akara Sucharitakul, Sheetal Patil (Sun Microsystems)
Recent papers (PDFs for most camera-ready papers can be found here):
- Above the Clouds: A Berkeley View of Cloud Computing. Written with many RAD Lab colleagues, our view of what’s really new about cloud computing, and its challenges and opportunities going forward.
- W. Sobel, S. Subramanyam, A. Sucharitakul, J. Nguyen, H. Wong, S. Patil, A. Fox, D. Patterson. Cloudstone: Multi-Platform, Multi-Language Benchmark and Measurement Tools for Web 2.0. Proc. 1st Workshop on Cloud Computing (CCA 08), Chicago, IL, October 2008.
More details:
In our CCA’08 paper we describe experiments to understand Web 2.0 app performance (both Rails and PHP versions of the app) on Amazon’s EC2. The links to download AMI’s to reproduce our results will appear here shortly, or you can email me. (Prove you’re not a bot by inferring my email address from this site!)