Indexing RETE's Working Memory: Catering to Dynamic Changes of the Ruleset

Simon Van De Water, Thierry Renaux, Lode Hoste, Wolfgang De Meuter

Research output: Unpublished contribution to conferenceUnpublished paper

Abstract

Complex Event Processing systems are used to detect patterns in large streams of events. These patterns are described by rules. Due to changing requirements or updated security policies for example, these rules need to be frequently updated at runtime. Changing the ruleset at runtime, however, is a costly operation because the events that occurred in the past need to be re-evaluated. In this paper we present an extension to the pattern-matching algorithm called RETE. Our extension aims to improve the performance of the re-evaluation of events when the ruleset is modified. More specifically, we optimise the addition of rules that match events by means of the inequality operators >, ≥, < and ≤. We do this by introducing indexes to order the events that are stored in the working memory. Afterwards, we measure our performance improvements on targeted micro-benchmarks.
Original languageEnglish
Number of pages7
Publication statusPublished - 7 Oct 2015
EventREBLS 2015: Reactive and Event-based Languages & Systems - Sheraton Station Square Hotel, Pittsburgh, United States
Duration: 27 Oct 201527 Oct 2015

Workshop

WorkshopREBLS 2015: Reactive and Event-based Languages & Systems
CountryUnited States
CityPittsburgh
Period27/10/1527/10/15

Keywords

  • Rete-Engine
  • indexing
  • Dynamic Ruleset

Fingerprint

Dive into the research topics of 'Indexing RETE's Working Memory: Catering to Dynamic Changes of the Ruleset'. Together they form a unique fingerprint.

Cite this