VLDB2020
BitFun: Fast Answers to Queries with Tunable Functions in Geospatial Array DBMS
Ramon Antonio Rodriges Zalipynis
被引用 2 次
摘要
Geospatial array DBMSs handle big georeferenced arrays. Due to the geospatial data peculiarities, many queries have tunable parameters with values not known in advance: users gradually tune them until they get a satisfactory result. This generates a series of queries with slightly different structures and very similar outputs. Modern array DBMSs spend the same efforts to answer each such query. BitFun provides novel bitmap indexing strategies to continuously re-index arrays during queries with similar mathematical functions. It can be up to 8x faster than computing the results from scratch. We describe BitFun and offer lessons on real-world geospatial data, related to real practical tasks. A lesson involves tuning a math function parameter while the rich web GUI details the indexing process and query execution. Conference attendees will appreciate BitFun approaches, its performance, and learn its internals via fascinating lessons.