Bayeux: an architecture for scalable and fault-tolerant wide-area data dissemination
NOSSDAV '01 Proceedings of the 11th international workshop on Network and operating systems support for digital audio and video
The Case for Resilient Overlay Networks
HOTOS '01 Proceedings of the Eighth Workshop on Hot Topics in Operating Systems
BRITE: An Approach to Universal Topology Generation
MASCOTS '01 Proceedings of the Ninth International Symposium in Modeling, Analysis and Simulation of Computer and Telecommunication Systems
Tapestry: An Infrastructure for Fault-tolerant Wide-area Location and
Tapestry: An Infrastructure for Fault-tolerant Wide-area Location and
Underlay-aware overlay networks
Underlay-aware overlay networks
Generalised Repair for Overlay Networks
SRDS '06 Proceedings of the 25th IEEE Symposium on Reliable Distributed Systems
ICDCSW '07 Proceedings of the 27th International Conference on Distributed Computing Systems Workshops
Availability models for underlay aware overlay networks
Proceedings of the second international conference on Distributed event-based systems
A Taxonomy of QoS-Aware, Adaptive Event-Dissemination Middleware
IEEE Internet Computing
Self-stabilizing publish/subscribe systems: algorithms and evaluation
Euro-Par'05 Proceedings of the 11th international Euro-Par conference on Parallel Processing
Hi-index | 0.00 |
Modern distributed applications that run on P2P overlays need the overlay to be resilient to failures in the underlying communications network. An example application is that of event dissemination where the P2P overlay (the event broker network) delivers events published by a client to subscribers based on subscription filters. Achieving high availability requires that the overlay be aware of, and utilize the path redundancies in the underlying physical network. In addition, the overlay should be self organizing, as broker nodes may dynamically join or leave the network and centralized control is not an option in large scale networks. In this paper, we present Trimarg, an efficient distributed algorithm for achieving a self organizing overlay with an availability degree of three. Our algorithm is based on a graph theoretic foundation for highly available overlay networks and is designed to handle the concurrency issues of a large scale distributed system. The resulting unstructured P2P overlay ensures 3-degree of availability in the presence of node and link failures in the underlying physical network. We have proved the correctness of the algorithm and analyzed its complexity to show that the time complexity is O (diameter *degree )2 of the network and the message complexity is O (diameter *degree ). Our algorithm is the first of its kind to deal with concurrency issues and self organizing capabilities of the resultant overlay.