Speakman, T., Farinacci, D., Lin, S., Tweedly, A., et al., Pragmatic General Multicast (PGM) Transport Protocol Specification, draft-speakman-pgrn-spec-04.txt, April 2000.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....However, there are certain known methods, such as delayed NACKing, that are able to minimize the undesired phenomena in the cost of slightly enlarged recovery latencies. Another option is to wait for additional M packets with SNs higher than the assumed loss before triggering NACK generation [34]. This method should be complemented with a timeout based mechanism that handles the loss of the last packet before a pause in the transmission of the data stream. 3.8 Deployment, Interoperability and Feasibility Figure 8 depicts a simplified modification of a forwarding path in an SDMP router. ....

....the router waits for the repair up to 10 seconds, after that it removes the state) When loss probability increases, RDATA losses become more dominant and contribute to the steep slope in the arrival latency graphs for PGM. A partial workaround for this weakness is proposed in a new PGM draft [34]. 2500 1500 500 Figure 11 Average packet arrival latency vs. link loss probability 15 140 130 10 120 5 100 90 1.0 0.0 0.2 0.4 0.6 0.8 ( Enlarged poruon ot igure 3) 0.95 0.9 0.85 0.8 0.75 0.7 0.65 0.6 0.55 0.5 4.2.2 Maximum Arriva Latency Figures 13 ....

[Article contains additional citation context not shown here]

Speakman, T., Farinacci, D., Lin, S., Tweedly, A., et al., Pragmatic General Multicast (PGM) Transport Protocol Specification, draft-speakman-pgrn-spec-04.txt, April 2000.

....polling of receivers to avoid the implosion problem [9] However, the receiver initiated approach is more dominant in the design of contemporary protocols because of its scalability advantages over the sender initiated approach. To name a few examples, SRM[ 10] RMTP[ 11 ] TMTP[ 12] and PGM[ 13] are all receiver initiated protocols that build relaibility on top of a best effort service. Further classification of reliable multicasting schemes is based upon the method used to recover from packet losses [14] Some protocols, for instance RAMP [25] use centralized error recovery, also ....

....NACKs for the same lost data) In order for these methods to work well, feedbacks should be multicast so they can be heard by other receivers. An accurate tuning of the timeout period is crucial, and is typically derived from RTT (Round Trip Time) estimates between participating nodes [ 10] [ 13 ] [ 19] Use of probabilistic feedback suppression methods impacts feedback latency and consequently recovery latency, as there is a tradeoff between the amount of feedback generated and the timeliness of the protocol. For example, when using a NACK random back off timer to supress feedbacks, each ....

[Article contains additional citation context not shown here]

Speakman, T., Farinaeel, D., Lin, S., Tweedly, A., Pragmatic GeneralMulticast (PGM) Transport Protocol Specification, in Internet Draft, August 1998.

....meaning the router waits for the repair up to 10 seconds, after that it removes the state) When loss probability increases, RDATA losses become more dominant and contribute to the steep slope in the arrival latency graphs for PGM. A workaround for this weakness is proposed in a new PGM draft [34] and is yet to be tested. Maximum arrival latency. Figures 8 and 9 show the effects of the dangling NAK state on the maximum arrival latency experienced by receivers. For each test run, the list of maximum arrival latencies experienced by all 64 receivers is averaged to a single value. This run ....

Speakman, T., Farinacci, D., Lin, S., Tweedly, A., et al., Pragmatic General Multicast (PGM) Transport Protocol Specification, draft-speakman-pgm-spec-04.txt, April 2000.

....approach does not decrease the amount of control messages that maintain the multicast forwarding state. At any time, routers still need to process these periodic messages on all multicast addresses in use concurrently. Recent proposals on router assisted multicast forwarding services, such as PGM [33], Breadcrumb Forwarding Services [39] and AIM [18] provide mechanisms to send data on a per packet basis to a subset of receivers on a multicast tree. Clustering is independent of the underlying multicast routing infrastructure, and is orthogonal to and works with these new services. ....

Speakman, T., Farinacci, D., Lin, S., and Tweedly, A. Pretty Good Multicast (PGM) Transport Protocol Specification, Jan. 1998. Internet Draft (RFC pending).

....packets sent to a multicast address are delivered to all end hosts subscribed to that address. Clustering is also independent of the underlying multicast routing infrastructure. It is orthogonal to and works with recent proposals on router assisted multicast forwarding services, such as PGM [32], Breadcrumb Forwarding Services [38] and AIM [18] These services provide mechanisms to send data, on a per packet basis, to a subset of receivers on a multicast tree. The amount of state and processing required at the routers is associated with the granularity of forwarding. We can apply ....

....correlates to topology, a hybrid scheme is clustering is necessary otherwise. The authors conclude that low latency address allocation and group creation should be supported by the Internet architecture to realize large scale multicast applications. Router level forwarding services such as PGM [32], Breadcrumbs [38] and GMTS [7] provide filtering mechanisms in multicast communication. Clustering can be incorporated into such services to prevent explosion of forwarding state and control overhead, especially when fine grained filtering is required. The Destination Set Group (DSG) scheme ....

Speakman, T., Farinacci, D., Lin, S., and Tweedly, A. Pretty Good Multicast (PGM) Transport Protocol Specification, Jan. 1998. Internet Draft (RFC pending).

....(SNAP) as part of the MASH [17] toolkit. We discuss some avenues for future work in this area. Global recovery in SRM results in request and reply floods transmitted to the entire group, even for losses that are localized. Several solutions have been proposed for local recovery in the literature [15, 7, 20, 27]. Integrating SNAP with a local recovery scheme could provide greater scalability. Finally, the deployment of IP multicast in the Internet has been impeded to some extent by the absence of a multicast congestion control algorithm. This problem has recently received significant attention in the ....

Speakman, T., Farinacci, D., Lin, S., and Tweedly, A. Pretty Good Multicast (PGM) Transport Protocol Specification, Jan. 1998. Internet Draft (RFC pending).

No context found.

T. Speakman, D. Farinacci, S. Lin, A. Tweedly, Pretty Good Multicast(PGM) Transport Protocol Specification, Internet Draft, January 1998

....only end system nodes to construct this hierarchy and attempts to automate the process using monitored loss rates between receivers. However, the stability of such a dynamic system is unstudied under realistic network conditions. 2.4. 3 Router assisted Schemes Pragmatic General Multicast (PGM) [127] is a combined network and transport layer solution to the problem of scalable wide area multicast. End hosts generate NACKs in response to missing packets. Routers are maintain transport level state that assists in suppressing duplicate NACKs from downstream members. Light weight Multicast ....

....(SNAP) as part of the MASH [82] toolkit. We discuss some avenues for future work in this area. Global recovery in SRM results in request and reply floods transmitted to the entire group, even for losses that are localized. Several solutions have been proposed for local recovery in the literature [74, 40, 98, 127]. Integrating SNAP with a local recovery scheme could provide greater scalability. Finally, the deployment of IP multicast in the Internet has been impeded to some extent by the absence of a multicast congestion control algorithm. This problem has recently received significant attention in the ....

Tony Speakman et al. Pragmatic Good Multicast (PGM) Transport Protocol Specification, June 1999. Internet Draft (RFC pending).

....in order to make optimal decision for all. However, it will rely on exchange preference messages among all users which will not scale well to large number of users. All above are end to end solutions which required no router support. Another set of approaches, such as subcast [12] AIM[6] PGM[15], etc. rely on some support from special routers. For example, routers above the bottleneck link will forward repair packets only to certain interfaces to reach users behind the bottleneck link. With special router support, it would be easy to identify the bottleneck links for efficient error ....

Tony Speakman, Dino Farinacci, Steven Lin, and Alex Tweedly. Pretty Good Multicast (PGM) Transport Protocol Specification. Cisco Systems, January 1998. Internet Draft.

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