Daniel Zappala, Deborah Estrin, and Scott Shenker, "Alternate Path Routing and Pinning for Interdomain Multicast Routing," Tech. Rep. USC-CS97 -655, Computer Science Department, University of Southern California, June 1997.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....achieves some degree of load balancing. On the other hand, traffic is more concentrated in CSP computation networks. To cope with this shortage in CSP, appropriate link cost functions that consider the available bandwidth of the link should be chosen. Studies regarding this issue can be found in [15]. In Fig. 2b, we have insight into the blocking probability, blocked in either the routing phase (prefix R ) or the signaling phase (prefix S ) Look at the performance under accurate network state (i.e. period = 0) the routing blocks account for all blocked flow requests. As the update ....

D. Zappala, D. Estrin, and S. Shenker, "Alternate Path Routing and Pinning for Interdomain Multicast Routing, " Tech. rep. 97--655, USC, 1997.

....of the hop plot exponent, H, for h , is c N 1; h 0 where c = N 2 E to satisfy initial conditions. The average neighborhood is a commonly used parameter in the performance of network protocols. Our estimate is an improvement over the commonly used estimate that uses the average degree [59] [51] which we call averagedegree estimate: NN (h) d ( d 1) h 1 In gure 7, we plot the actual and the two estimates of the average neighborhood size versus the number of 7 hops using an instance from 1998. The superiority of the hop plot exponent estimate is apparent compared to the ....

D. Zappala, D. Estrin, and S. Shenker. Alternate path routing and pinning for interdomain multicast routing. Technical Report USC CS TR 97-655, U. of South California, 1997.

....of the multicast trees is analyzed. Group dynamics is added, and the impact on tree cost and feasibility is studied. The main problem is that when a member is added or pruned from a tree, the resulting tree may not be the minimum cost tree for the new con guration. Alternate path routing MORF [104] is a multicast setup protocol intended to be used in conjunction with the multicast routing protocol. The basic idea of MORF is to provide alternate paths for applications, in the case where the opportunistic default route is not able to support the QoS requirements; and to allow the possibility ....

D. Zappala and D. Estrin, Alternate Path Routing and Pinning for Interdomain Multicast Routing. USC, 1997. Computer Science Technical Report.

....based on shortest paths. To achieve a desired bandwidth, latency, Internet Draft, PGM Reliable Transport Protocol Specification, D. Farinacci et al. January 1998. 21 or other quality of service target, however, alternative paths may be indicated. A scheme for constructing them is described in [ Zappala et al. 1997 ] ffl QOS forwarding. Once a reservation is made along a network path, it is desirable that traffic follow that path. This may not occur with IP routing, since each packet is independently forwarded along the current shortest path. To address this mismatch, new forwarding services, for instance ....

Daniel Zappala, Deborah Estrin, and Scott Shenker. Alternate Path Routing and Pinning for Interdomain Multicast Routing. Technical Report TR97-655, CS Dept., University of Southern California, 1997.

....approach of exchanging dynamic link state metrics to compute QoS based routes does not scale to large networks. We can distinguish two categories among Internet multicast protocols based on QoS considerations: QoS oblivious protocols [1] 10] 19] 17] and QoS sensitive protocols [3] [25]. An overview of the above protocols is presented in the next section. Our protocol was designed with the following primary goals. ffl QoS Support. We want to provide a framework to support arbitrary QoS requirements of users. To achieve this, we have to consider multiple paths, and handle the ....

....design. In Appendix A, we present an extended list of the efficiency aspects of protocols. In our protocol, we compare paths in terms of their ability to support an application at a specific QoS level. Quality of Service (QoS) denotes the user perceived quality. Recently, Zappala et al. [25] used the term Quality of Route (QoR) to refer to multiple static parameters of the route (e.g. link capacity, delay, or reliability) and this was adopted in the YAM protocol [3] In our work, we suggest the use of dynamic metrics, e.g. available bandwidth, current delay) because these provide ....

[Article contains additional citation context not shown here]

D. Zappala, D. Estrin, and S. Shenker. Alternate path routing and pinning for interdomain multicast routing. Technical Report USC CS TR 97-655, U. of South California, 1997.

....approach of exchanging dynamic link state metrics to compute QoS based routes does not scale to large networks. We can distinguish two categories among Internet multicast protocols based on QoS considerations: QoS oblivious protocols [1] 10] 18] 16] and QoS sensitive protocols [4] [24]. An overview of the above protocols is presented in the next sec tion. Our protocol was designed with the following primary goals. ffl QoS Support. We want to provide a framework to support arbitrary QoS requirements of users. To achieve this, we have to consider multiple paths, and handle ....

....properties to large networks is an overriding concern for protocol and network architecture design. In our protocol, we compare paths in terms of their ability to support an application at a specific QoS level. Quality of Service (QoS) denotes the user perceived quality. Recently, Zappala et al. [24] used the term Quality of Route (QoR) to refer to multiple static parameters of the route (e.g. link capacity, delay, or reliability) and this was 2 Source Based Trees require a routing entry per source per group, while Shared Trees require a single routing entry per group. adopted in the YAM ....

[Article contains additional citation context not shown here]

D. Zappala, D. Estrin, and S. Shenker. Alternate path routing and pinning for interdomain multicast routing. Technical Report USC CS TR 97-655, U. of South California, 1997.

....exponent, H, for h ffi, is NN(h) c N h H Gamma 1 where c = N 2 E to satisfy initial conditions. The average neighborhood is a commonly used parameter in the performance of network protocols. Our estimate is an improvement over the commonly used estimate that uses the average outdegree [26] [7] which we call averageoutdegree estimate: NN 0 (h) d ( d Gamma 1) h Gamma1 In figure 9, we plot the actual and both estimates of the average neighborhood size versus the number of hops for the Int 12 98 graph. In Table 3, we show the normalized error of each estimate: we calculate ....

D. Zappala, D. Estrin, and S. Shenker. Alternate path routing and pinning for interdomain multicast routing. Technical Report USC CS TR 97-655, U. of South California, 1997.

No context found.

Daniel Zappala, Deborah Estrin, and Scott Shenker, "Alternate Path Routing and Pinning for Interdomain Multicast Routing," Tech. Rep. USC-CS97 -655, Computer Science Department, University of Southern California, June 1997.

....III B.3) We have chosen to detect and break loops in this manner with APM because the join latency is small in the common case when there is no loop formed. In this case, the latency is only the time it takes to traverse a source route from the receiver to the multicast tree. In prior work [27], 23] we designed a setup protocol with the opposite tradeoff, namely loop prevention at the expense of larger latency in the common case. Since formation of a loop requires several receivers trying to use an alternate path at the same time, with routes that overlap to form a loop, and with ....

Daniel Zappala, Deborah Estrin, and Scott Shenker, "Alternate Path Routing and Pinning for Interdomain Multicast Routing," Tech. Rep. USC-CS-97-655, Computer Science Department, University of Southern California, June 1997.

....III B.3) We have chosen to detect and break loops in this manner with APM because the join latency is small in the common case when there is no loop formed. In this case, the latency is only the time it takes to traverse a source route from the receiver to the multicast tree. In prior work [29], 25] we designed a setup protocol with the opposite tradeoff, namely loop prevention at the expense of larger latency in the common case. Since formation of a loop requires several receivers trying to use an alternate path at the same time, with routes that overlap to form a loop, and with ....

D. Zappala, D. Estrin, and S. Shenker, "Alternate Path Routing and Pinning for Interdomain Multicast Routing," Tech. Rep. USC-CS-97-655, University of Southern California, June 1997.

No context found.

Zappala, D., Estrin, D., and S. Shenker, "Alternate Path Routing and Pinning for Interdomain Multicast Routing", USC Computer Science Technical Report #97-655, USC, 1997.

No context found.

D. Zappala, D. Estrin, and S. Shenker. Alternate path routing and pinning for interdomain multicast routing. Technical Report USC CS TR 97-655, U. of South California, 1997. 124

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