Harnessing the Clouds: From Systems to Applications
The last few years have seen explosive growth in industries providing data storage coupled with data processing. Basically, a company like Amazon or Google offers services to customers through a computer cluster (servers+network) that combines large-scale data storage with a co-located computing facility. Such a system has come to be known as a cloud, and there are many questions about how to design and deploy an optimal Cloud Operating System.
Professor Gupta is co-principal investigator at the University’s new Cloud Computing Testbed, a collaboration with Hewlett-Packard, Yahoo!, Intel, and the National Science Foundation. During his Center appointment he will lead research in five areas at the forefront of cloud computing:
Cloud monitoring systems: Build systems that allow expressive, fast, and accurate queries to be executed across hundreds or thousands of servers; build a system for efficient central collection of large application logs to a data server.
Computation paradigms for large data: Tackle the twin problems of efficient data scheduling and efficient computation scheduling.
Crawling online social networks: Address questions of concurrently running crawlers and their data processing engines, the interface between the two, and how to pipeline data between these phases.
Cloud applications: Deploy application systems and improve and introduce new services to benefit application systems.
Closing the loop: Share knowledge gained by the Cloud Computing Testbed to benefit improved cloud computing.
These research areas are part of a larger effort to answer fundamental long-term questions: What services belong in a complete Cloud Operating System? What are the limits of the cloud-computing paradigm? How to best increase the scalability and performance of cloud systems, design new distributed algorithms for clouds, and encourage the design and development of new research applications atop clouds?