DBLab School of Computer and Electrical Engineering KDBSL NTUA
Thursday, July 02, 2020

Window Update Patterns in Stream Operators

Continuous queries applied over nonterminating data streams usually specify windows in order to obtain an evolving –yet restricted– set of tuples and thus provide timely results. Among other typical variants, sliding windows are mostly employed in stream processing engines and several advanced techniques have been suggested for their incremental evaluation. In this work, we set out to study the existence of monotonic-related semantics in windowing constructs towards a more efficient maintenance of their changing contents. We investigate update patterns observed in common window variants as well as their impact on windowed adaptations of typical operators (like selection, join or aggregation), offering more insight towards design and implementation of stream processing mechanisms. To demonstrate its significance, this framework is validated for several windowed operations against streaming datasets with simulations at diverse arrival rates and window sizes. Finally, we will discuss some possible extensions of this work, particularly towards static optimization with query rewriting rules involving windows.

[ Back ]