A Multi-way Semi-stream Join for a Near-Real-Time Data Warehouse

Abstract

Semi-stream processing, the operation of joining a stream of data with non-stream disk-based master data, is a crucial component of near real-time data warehousing. The requirements for semi-stream joins are fast, accurate processing and the ability to function well with limited memory. Currently, semi-stream algorithms presented in the literature such as MeshJoin, Semi-Stream Index Join and CacheJoin can join only one foreign key in the stream data with one table in the master data. However, it is quite likely that stream data have multiple foreign keys that need to join with multiple tables in the master data. We extend CacheJoin to form three new possibilities for multi-way semistream joins, namely Sequential, Semi-concurrent, and Concurrent joins. Initially, the new algorithms can join two foreign keys in the stream data with two tables in the master data. However, these algorithms can be easily generalized to join with any number of tables in the master data. We evaluated the performance of all three algorithms, and our results show that the semi-concurrent architecture performs best under the same scenario.

Semi-Stream Similarity Join Processing in a Distributed Environment

“…Content may change prior to final publication. [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16] Hadoop no Similarity [17], [18], [19] Spark no Similarity [20], [21], [22], [23], [24], [25], [26], [27] N/A yes Equi [28] N/A yes Similarity [29] Spark yes Equi DSim-Join Spark yes Similarity…”

“…Mehmood and Naeem [22] used parallel execution between the stream-probing phase and the disk-probing phase by introducing an intermediate buffer between the two phases. Naeem, Nguyen, and Weber [23] proposed multi-way semi-stream join methods. Naeem et al [24] presented a technique for load shedding in semistream join processing.…”