Tina Wong, Randy Katz, and Steven MacCanne. A preference clustering protocol for large-scale multicast applications. In Proceedings of the First International Workshop on Networked Group Communication, November 1999.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....on interests in application content. They conclude that a mixture of group addressing and ltering are an advantage when multicast addresses are scarce and an ecient mapping algorithm is used. The solution in this paper can be seen as a way to make ecient use of multicast groups. T. Wong et al. [17], present a clustering algorithm to group data sources and receivers, but show only the applicability to subject based. In the following section, the problem of mapping subscriptions into multicast groups is addressed. 3 The Multicast Mapping Problem This section addresses the mapping problem. ....

Tina Wong, Randy Katz, and Steven MacCanne. A preference clustering protocol for large-scale multicast applications. In Proceedings of the First International Workshop on Networked Group Communication, November 1999.

....services the application: all data reaches intended receivers, ignoring packet loss. It should be apparent that addressed applications more efficiently service the application. However, most applications and transport protocols rely on a broadcast model, and a growing set rely on a hybrid approach [30]. In the following section, we introduce a framework to discuss these design options. In the subsequent sections, we model and simulate the support choices to show the network characteristics of each scenario. Then, we discuss why the Internet is not prepared to support the addressing model though ....

....explosive algorithm, but within reasonable execution times. The tests we ran also showed that the mapping algorithm we used always performed better than random mapping. As we discuss subsequently, our results suggest the hybrid approach and the design of this kind of difficult mapping algorithm [30] should be avoided by instead designing a supportive multicast architecture. V. PERFORMANCE RESULTS We ran two major sets of simulations: one set for transaction applications, and one set for long duration applications. Within each set, the same parameters were used, and thus, the same sequence ....

T. Wong, R. Katz, and S. McCanne, "A preference clustering protocol for large-scale multicast applications," in Proceedings of the First International Workshop on Networked Group Communication, November 1999.

....to carry out multicast forwarding, and keeping membership state at the non branching routers is unnecessary. Application level clustering Large scale applications such as content delivery and distributed interactive simulation (DIS) use a significant number of multicast addresses. Clustering [11, 23, 24] exploits application level knowledge, such as users preferences in content or players positions in virtual space, to intelligently group sources and receivers into a limited number of multicast groups. 1.3 Our Contributions Although much research has focused on the reduction of multicast ....

T. Wong, R. Katz, and S. McCanne. A Preference Clustering Protocol for Large-Scale Multicast Applications. In Proceedings of Networked Group Communication Workshop, Pisa, Italy, November 1999.

....that . in the long run, the biggest issue facing multicast deployment is likely to be the scalability of multicast forwarding state as the number of multicast groups increases. 21] Several recent proposals reduce multicast state through aggregation [17, 21] application level clustering [14, 25, 26], tunneling [5, 6] and non branching state elimination [22, 20] We briefly describe these works next. 1.2 Multicast State Reduction State aggregation Forwarding state aggregation [21] replaces multiple forwarding entries with a single one, if the entries have adjacent group address prefixes ....

....to carry out multicast forwarding, and keeping membership state at the non branching routers is unnecessary. Application level clustering Large scale applications such as content delivery and distributed interactive simulation (DIS) use a significant number of multicast addresses. Clustering [14, 25, 26] exploits application level knowledge, such as users preferences in content or players positions in virtual space, to intelligently group sources and receivers into a limited number of multicast groups. 1.3 Our Contributions Although much research has been focused on the reduction of multicast ....

Wong, T., Katz, R., and McCanne, S. A Preference Clustering Protocol for Large-Scale Multicast Applications. In Proceedings of Networked Group Communication Workshop (Pisa, Italy, November 1999).

....and processing requirements. Our framework is generic in that an application can customize the algorithm according to its requirements and data characteristics. The protocol to coordinate sources and receivers to perform clustering is outside the scope of this paper, and we describe it elsewhere [21]. We conducted detailed simulation experiments to study various issues and tradeoffs in applying clustering to different preference patterns, data types, and application classes. We found that: ffl clustering successfully exploits preference similarity in multicast applications and utilizes ....

WONG, T., KATZ, R., AND MCCANNE, S. A Preference Clustering Protocol for Large-Scale Multicast Applications. In Proceedings of Networked Group Communication Workshop (Pisa, Italy, November 1999).

No context found.

T. Wong, R. Katz, and S. McCanne, "A preference clustering protocol for largescale multicast applications," in Networked Group Communication - First International COST26 Workshop, NGC'99, (Pisa, Italy), Nov. 1999.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC