Scrambling query plans to cope with unexpected delays

Accessing data from numerous widely distributed sources poses significant new challenges for query optimization and execution. Congestion and failures in the network can introduce highly variable response times for wide area data access. The paper is an initial exploration of solutions to this variability. We introduce a class of dynamic, run time query plan modification techniques that we call query plan scrambling. We present an algorithm that modifies execution plans on-the-fly in response to unexpected delays in obtaining initial requested tuples from remote sources. The algorithm both reschedules operators and introduces new operators into the query plan. We present simulation results that demonstrate how the technique effectively hides delays by performing other useful work while waiting for missing data to arrive