Title :
Improved concurrency control techniques for multi-dimensional index structures
Author :
Kanth, K. V Ravi ; Serena, David ; Singh, Ambuj K.
Author_Institution :
Dept. of Comput. Sci., California Univ., Santa Barbara, CA, USA
fDate :
30 Mar-3 Apr 1998
Abstract :
Multi-dimensional index structures such as R-trees enable fast searching in high-dimensional spaces. They differ from uni-dimensional structures in the following aspects: index regions in the tree may be modified during ordinary insert and delete operations; and node splits during inserts are quite expensive. Both these characteristics may lead to reduced concurrency of update and query operations. We examine how to achieve high concurrency for multi-dimensional structures. First, we develop a new technique for efficiently handling index region modifications. Then, we extend it to reduce/eliminate query blocking overheads during node-splits. We examine two variants of this extended scheme: one that reduces the blocking overhead for queries, and another that completely eliminates it. Experiments on image data on a shared-memory multiprocessor show that these schemes achieve up to 2 times higher throughput than existing techniques, and scale well with the number of processors
Keywords :
concurrency control; database theory; query processing; shared memory systems; software performance evaluation; tree data structures; visual databases; R-trees; concurrency control; delete operation; experiments; fast searching; high-dimensional spaces; image data; index region modifications; index regions; insert operation; multidimensional index structures; node splits; query blocking overheads; query operations; shared memory multiprocessor; update operations; Computer science; Concurrency control; Concurrent computing; Database systems; Image databases; Indexes; Indexing; Information retrieval; Spatial databases; Throughput;
Conference_Titel :
Parallel Processing Symposium, 1998. IPPS/SPDP 1998. Proceedings of the First Merged International ... and Symposium on Parallel and Distributed Processing 1998
Conference_Location :
Orlando, FL
Print_ISBN :
0-8186-8404-6
DOI :
10.1109/IPPS.1998.669984