Quasar: A Probabilistic Publish-Subscribe System

As describes by Bernard Wong Saikat Guha Dept. of Computer Science, Cornell University the comon peer-to-peer publish-subscribe systems rely on structured-overlays and rendezvous nodes to store and relay group membership information.  This design incurs the significant cost of creating and maintaining rigid-structures and introduces hotspots in the system at nodes that are neither publishers nor subscribers.

Memory Vaults is built on Quasar, a rendezvous-less probabilistic publish-subscribe system that caters to the specific needs of social networks. It is designed to handle networks of many groups creating a routing infrastructure based on the proactive dissemination of highly aggregated routing vectors to provide anycast-like directed walks in the overlay. This primitive, when coupled with a novel mechanism for dynamically negating routes, enables scalable and efficient group-multicast that obviates the need for structure and rendezvous nodes.  Memory Vaults current challenge is to proof that a a large-scale system is fficient.

Refrence: https://www.microsoft.com/en-us/research/wp-content/uploads/2008/02/iptps08-quasar.pdf