Study of concurrent operations on R-trees

[[abstract]]Appropriate concurrent operations can improve R-tree efficiency. Supporting concurrency control algorithms for the original R-tree is necessary since many database systems have adopted the original R-tree data structure as the basic spatial data access method. This paper presents concurrency control algorithms that do not alter the original R-tree structure for range searches, insertions, and deletions. The algorithms not only yield correct results, but also ensure deadlock freedom. Cost analyses and lock efficiency are analyzed mathematically. Algorithm performance is evaluated by using a simulation model. Simulation results show the R-tree performs more efficiency if run in a multidisk environment with a low fanout structure.[[fileno]]2030204010001[[department]]資訊工程學