Parallel Query Processing on the Grid

Database queries offer an easy-to-use declarative manner for describing complex data management tasks. Query processing technologies have been evolving for decades; however, the emergence of the Grid creates a new setting in which novel research issues and challenges have arisen. This paper discusses how Grid-oriented and/or service-based query processors differ from traditional ones, and focuses on three complementary research issues, namely, how to schedule parallel database queries over non-dedicated, distributed resources; how to mitigate the impact of increased data transfer cost; and how to perform load balancing in this new setting.