Diplomarbeit DIP-3150

Bibliograph.
Daten
Schieberle, Christian: Underlay aware approach to support quality of service in publish-subscribe systems.
Universität Stuttgart, Fakultät Informatik, Elektrotechnik und Informationstechnik, Diplomarbeit Nr. 3150 (2011).
73 Seiten, englisch.
CR-Klassif.C.2.4 (Distributed Systems)
C.2.2 (Network Protocols)
Kurzfassung

Providing delay-reduced routing is important in publish-subscribe systems where timely de- livery of event notifications is a critical factor affecting system operation or user experience. However, common research focused primarily on alleviating false-positives. More recent ef- forts aim towards quality related issues through adapting the overlay according to subscriber requirements but leaving underlying network characteristics aside.

It is commonly accepted that efficient routing can only be achieved when underlying network characteristics are respected. Even so, incorporating underlay-aware strategies to build low- stretch overlays is not considered in many distributed environments.

This work focuses on solving the problem of establishing an efficient underlay-aware routing mechanism in a content-based publish-subscribe system. In particular, we strive to reduce end-to-end delay among communication partners. Thereby, our contributions are twofold: We will develop a topology inference scheme for unstructured peer-to-peer networks and introduce a routing mechanism reducing overall end-to-end delay among peers. Experimental evaluations will be given for different Internet-like router topologies showing that the approach is capable of modeling an underlay network in an efficient and accurate manner. Furthermore, we will show the positive impact on the stretch of the overlay to outline the concept as a source for efficient event notification delivery in a publish-subscribe environment.

Volltext und
andere Links
PDF (1884954 Bytes)
Abteilung(en)Universität Stuttgart, Institut für Parallele und Verteilte Systeme, Parallele Systeme
BetreuerM. Sc. M. Adnan Tariq
Eingabedatum13. September 2011
   Publ. Institut   Publ. Informatik