Title :
Batched Processing for Information Filters
Author :
Fischer, Peter M. ; Kossmann, Donald
Author_Institution :
Swiss Fed. Inst. of Technol., Zurich, Switzerland
Abstract :
This paper describes batching, a novel technique in order to improve the throughput of an information filter (e.g. message broker or publish & subscribe system). Rather than processing each message individually, incoming messages are reordered, grouped and a whole group of similar messages is processed. This paper presents alternative strategies to do batching. Extensive performance experiments are conducted on those strategies in order to compare their tradeoffs.
Keywords :
information filtering; information filters; batched processing; information filter; message broker; subscribe system; Connectors; Databases; Delay; Indexing; Information filtering; Information filters; Query processing; Scalability; Throughput; XML;
Conference_Titel :
Data Engineering, 2005. ICDE 2005. Proceedings. 21st International Conference on
Print_ISBN :
0-7695-2285-8
DOI :
10.1109/ICDE.2005.25