Interactive query processing partial query results dynamic user control during query execution adaptive query execution interactive data cleaning and transformation
achieve interactive behavior by adapting the query execution changing user interests changing data source properties Adaptive query execution S T R R S T R S T T S T R traditional query plan fine-granularity adaptation adaptive tuple ordering adaptive tuple and operator ordering
relational operators: logical abstractions –encapsulate multiple physical effects –inflexible in handling unexpected changes cannot gracefully adapt –access method/data source selection –join algorithm selection –resource allocation: e.g. memory –delays want to encapsulate at level of physical operators Granularity of Query Operators R Eddy P ind.jn hash jn S P RS
State Modules isolate (and share) state in State Modules (SteMs) work sharing query execution = routing thru sources, SteMs can adapt access methods, join algorithms gracefully directly measure & adapt resource consumption P R Eddy P ind.j n hash jn R R Eddy P P