Efficient Sliding Window Join in Data Stream Processing

Our previous work compared two possible approaches to process sliding window join over continuous data streams. It showed that using multiple hash tables per stream source provided better performance than using a single hash table. On the other hand, performance of the single-table scheme can be improved by reducing the windowing cost to deal with expired tuples. In this paper, we discuss how to reduce the cost in the single-table scheme algorithm as a solution for efficient sliding window join.