L. wei H. Lehman, S. J. Garland, and D. L. Tennenhouse. Active reliable multicast. In Proceedings of IEEE INFOCOM '98, 1998.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....several previously proposed services. 1 Introduction Several proposals have emerged in recent years for the deployment of active services in routers to support multicast transport. In the earliest proposals, the augmented router functionality was targeted at facilitating reliable multicast [6, 8, 13, 11, 15, 7], a problem that has resisted a scalable solution using purely endto end techniques. Other work [17] promotes active filtering services to improve the performance of specific applications. Recent efforts [3, 4] have sought to generalize previous work by defining a simple set of services (feedback ....

....the forward multicast path. Periodic multicast of fresh SPMs allows the signaling path to adapt to changes in the underlying tree and, since parent addresses are stored as softstate, unused branches to be pruned automatically. Other network supported reliable multicast schemes have been proposed [8, 13, 11, 15, 7]. Space does not permit us to review them all here. It is worth observing, however, that LMS [10] differs significantly from PGM by enlisting the network to support a hierarchical approach. PGM and LMS offer different trade offs between storage requirements at active nodes and the degree to which ....

Li wei H. Lehman, Stephen J. Garland, and David L. Tennenhouse. Active reliable multicast. In Proc. Infocom'98, 1998.

....Patterns Analyzing deployment requirements of services for active networks, we observed patterns that guide the deployment. When classified according to deployment information flow, different services may be ranged in the same category. For example, comparing a reliable multicast service [10] and a video scaling service [6] one notices that in both cases the service components have to be deployed on the routers located at the nodes of a multicast tree. The main difference between the two services is the actual logic of the service components. Thus service deployment must only deal ....

Li wei Lehman, Stephen J. Garland, and David L. Tennenhouse. Active reliable multicast. In Proceedings of IEEE Infocom, San Francisco, USA, 1998. 11

....Most proposals sketch incremental deployment plans, but very few carry out systematic evaluations. In addition to the earlier work on incremental deployment of LMS [4] described in Section I, the other related work we are aware of is a study in the context of Active Reliable Multicast(ARM) [10]. ARM argues that significant benefits 200 400 600 800 1000 1200 1400 0 2 4 6 8 10 Maximum Peak NAKs Back Off Timer Interval (unit: RTT) PGM: Maximum Peak NAKs Fig. 22. PGM back off timer interval sensitivity: PGM Maximum Peak NAKs with different back off timer intervals 0 20 40 ....

L. wei Lehman, S. J. Garland, and D. L. Tennenhouse, "Active Reliable Multicast," in Proceedings of the IEEE Infocom'98, San Francisco, USA, March 1998.

....by executing application dependent functions on incoming packets. Recently, the use of active network concepts where routers could contribute to enhance the network services by customized (sometimes high level) functionalities has been proposed in many research areas including multicast protocols [11, 12] and distributed interactive simulations [13] These results can be very bene cial to the grid community on numerous ways: i) program and data remote submissions, ii) information services and naming services, iii) data replication and caching, iv) collective operations and clock ....

....for local recoveries of loss packets and reduces the recovery latency. the suppression of NACKs reduces the NACK implosion problem. the subcast of repair packets to a set of receivers limits both the retransmission scope and the bandwidth usage. For instance, ARM (Active Reliable Multicast) [11] and AER (Active Error Recovery) 12] are two protocols that use a best e ort cache of data packets to permit local recoveries. ARM adopts a global suppression strategy: a receiver experiencing a packet loss sends immediately a NACK to the source. Active services in routers then consist in the ....

L. Wei, H. Lehman, S. J. Garland, and D. L. Tennenhouse. Active reliable multicast. In IEEE INFOCOM'98.

....play an active role by executing applicationdependent functions on incoming packets. Recently, the use of active network concepts [21] where routers themselves could contribute to enhance the network services by customized functionalities have been proposed in the multicast research community [22]. Contributing mainly on feedback explosion problems, retransmission scoping and cache of data, these active reliable multicast protocols open new perspectives for achieving high throughput and low latency on wide area networks. Caching data packets allows for local recoveries of lost packets and ....

L. Wei, H. Lehman, S. J. Garland, and D. L. Tennenhouse. Active reliable multicast. In IEEE INFOCOM'98, March 1998.

.... In this perspective, a number of execution environments, most of them based on the Java language, have been proposed to allow the dynamic execution of user code [14, 4] Recently, the use of these active network concepts has been proposed in many research areas including multicast protocols [15, 6, 1], distributed interactive simulations [18] and network management [11] There are many difficulties, however, in deploying in a large scale an active networking infrastructure. Security and performance are amongst these difficulties. However, active networking has the ability to provide a very ....

....active solutions can provide efficient mechanisms to the reliability problem, and especially to enabling fast local recoveries. Most active services proposed so far in multicast protocols contribute mainly on feedback implosion problems, retransmission scoping and cache of data. For instance, ARM [15], AER [6] and RMANP [1] are protocols that use cache of data packets to permit local recoveries and advanced NACK suppression strategies. 3 The DyRAM Framework 3.1 Overview DyRAM [7] is a reliable multicast protocol suite with a recovery strategy based on a tree structure constructed on a ....

L. Wei, H. Lehman, S. J. Garland, and D. L. Tennenhouse. Active reliable multicast. In IEEE INFOCOM '98.

....Most proposals sketch incremental deployment plans, but very few carry out systematic evaluations. In addition to the earlier work on incremental deployment of LMS [4] described in Section I, the other related work we are aware of is a study in the context of Active Reliable Multicast(ARM) [10]. ARM argues that significant benefits can be obtained even when only 50 of the routers are ARMcapable. Further, the authors suggest that significant benefits can be obtained even with a much smaller set of ARM capable routers if strategically located, but they do not investigate what these ....

L. wei Lehman, S. J. Garland, and D. L. Tennenhouse, "Active Reliable Multicast," in Proceedings of the IEEE Infocom'98, San Francisco, USA, March 1998.

....[6] and Tracer [12] use the help of the mtrace [19] utility to build the hierarchy. These schemes require support from the routers in order to operate correctly and efficiently. The only two works we are aware of that study the impact of incremental deployment for reliable multicast are [14] and [20]. The first one investigates the performance of LMS under various deployment schemes; however it does not consider PGM or any other schemes. The topologies for that study are generated by the GT ITM [16] topology generator and are quite small (around 400 nodes) In this paper we use a real ....

....(around 400 nodes) In this paper we use a real Internet router level topology with 27646 nodes. This topology is bigger and more realistic than the generated topologies, and it also contains AS information so that we can study deployment schemes related with AS. Active Reliable Multicast(ARM) [20] utilizes soft state storage within the network for NACK suppression and to limit the data retransmission only to receivers that have observed losses. Incremental deployment strategies are studied within the context of ARM. The authors have found that significant benefits can be obtained when only ....

Li wei Lehman, Stephen J. Garland, and David L. Tennenhouse, "Active Reliable Multicast," in Proceedings of the IEEE Infocom '98, San Francisco, USA, March 1998.

....in processed bandwidth occurs. It may also be seen that at very low levels of bu ering, the amount of data that can be processed drops o rapidly due to the inherent latency in the Nemesis I O channels between the network device driver and Rcane. D. Memory Isolation D. 1 Memory Consumption In [LGT98] the Active Reliable Multicast protocol (ARM) was proposed to improve the performance of reliable multicast ows. ARM caches packets at intermediate nodes within the network, trading memory on the nodes for reduced bandwidth requirements and latency; as such, some form of resource control is ....

Li-wei Lehman, Stephen J. Garland, and David L. Tennenhouse. Active Reliable Multicast. In Proceedings of IEEE Infocom '98, March 1998.

.... provided by the active router to shorten TCP start up time and optimize the rate control, which can save much bandwidth from the many short TCP connections [74] In MIT Active Networks project, the authors showed that active nodes and protocols could be used to control congestion for multicast [50], and improve the performance of distributed applications [49] In the University of Kansas Magican project, active network is used to implement a congestion control algorithm for real time voice packets traversing a low bandwith ATM link [78] Page 11 4. Node to Node Congestion Control The ....

Li-wei Lehman, Stephen J. Garland, and David L. Tennenhouse, Active Reliable Multicast, IEEE INFOCOM'98.

....assumptions so that an empirical analysis of our system will yield results that closely emulate real world scenarios. o oo A wide range of work in the development of end user application services for active networks is currently being done. This includes Active Reliable Multicast [25], improvements in network caching [15, 32] Network Security [10] Active Bridging [12] data fission and fusion techniques within the network [37] and application oriented congestion control mechanisms [30, 33, 35] In future implementations we wish to take a closer look at userlevel ....

Li-wei Lehman, Stephen J. Garland, and David L. Tennenhouse, Active Reliable Multicast. IEEE INFOCOM `98, San Fransisco, CA, April 1998.

....approaches is the use of active packets (or capsules, in the terminology of [30] In an active packet, the traditional packet header is replaced by a program that determines the packet s handling in the network. Active packets have enabled various applications ranging from reliable multicast [35] to auction servers [37] to application directed routing [13] Although there have been a variety of active packet system designs and there are several publicly available implementations, the idea has not taken substantial hold outside of the core AN community. We argue that, to date, none of the ....

....performance adequate to T1 line rates. Flexibility. Perhaps the best demonstration of the flexibility of the ANTS system is the number and variety of active networking applications which have been built using the ANTS toolkit. These include specialized web caching [15] reliable multicast [35], and distributed auction services [16] ANTS nodes may be scattered throughout the network; legacy routes may just perform default IP forwarding, thus permitting incremental ANTS deployment. One scalability drawback is that the MD5 fingerprint of the protocol code makes capsule code re use ....

Li wei H. Lehman, Stephen J. Garland, and David L. Tennenhouse. Active reliable multicast. In IEEE INFOCOM'98, March/April 1998.

....of disjoint multicast groups where the hierarchy is congruent with the multicast tree topology from the source of the session. Other researchers have proposed the use of intelligent computing in the network to assist in the design of reliable multicast protocols. Active Reliable Multicast (ARM) [44] uses the concept of active routers that can perform customized computation on behalf of the end points. They provide best effort softstate storage and perform data caching for local retransmission, NACK fusion suppression, and partial multicasting for scoped retransmission. IX. SUMMARY We ....

Li-Wei H. Lehman, Stephen J. Garland, and David L. Tennenhouse, "Active Reliable Multicast," in Proceedings of INFOCOM '98, Mar. 1998.

....duplication of packets and immunity to group membership changes, 7 while existing passive schemes provide only partial solutions to the above problems. Indeed, at MIT the inventors of a loss recovery scheme that takes advantage of active networking, called active reliable multicast (ARM) [18], claim that it can solve all the above problems efficiently. Active Reliable Multicast and Other Research ARM utilizes intermediate routers to protect the sender and network bandwidth from unnecessary acknowledgments and retransmissions. ARM routers play an active role in loss recovery. They ....

Li-wei H. Lehman, Stephen J. Garland, and D.L. Tennenhouse, Active Reliable Multicast, Proc. IEEE INFOCOM `98, San Francisco, CA, 29 March--2 April 1998.

.... model, where all packets are regarded as programs[TSS 97] The research has now produced a variety of results, including working infrastructures such as ANTS[WGT98] PLAN[HKM ] and Alien[ASNS97] and applications of these infrastructures to problems ranging from reliable multicast[wLGT98] and auctions[LWG98] to more traditional networking tasks such as error control[FMS 98] congestion control and bridging[ASNS97] It has become clear that a general architecture for active network elements can be defined using a separation of concerns. Applications for the switch model, the ....

....and integrity of the packet must be ensured. The digital signature that binds a credential to a packet al..so provides this protection if the credential is associated with the source of the packet. 6 Novel Applications of Active Networks 6. 1 Active Reliable Multicast Active Reliable Multicast[wLGT98] ARM) embeds active routers in a multicast tree. The routers differ from conventional routers in their handling of loss recovery in a reliable multicast setting. The availability of processing and storage can also be used to enhance performance. The central ideas in the loss recovery improvement ....

Li wei Lehman, Stephen J. Garland, and David L. Tennenhouse. Active Reliable Multicast. In Proceedings, IEEE INFOCOM 1998, pages 581--589, April 1998.

....issues that need to be addressed. Section 7 explains two alternative implementations. We conclude the paper with results from our simulations. 2 Related Work The problem of feedback implosion has been addressed in the context of reliable multicast. For example, many reliable multicast schemes [17, 42, 30, 35] propose suppressing NACKs to solve the problem of NACK implosion. In such schemes, if a node within the network receives identical NACKs from multiple nodes, it forwards just one NACK to the source. Similarly, the developers of the Cu SeeMe system, from Cornell University, advocate the concept of ....

Li wei H. Lehman, Stephen J. Garland, and David L. Tennenhouse. Active reliable multicast. In Proceedings of the IEEE INFOCOM, pages 581--589, March 1998.

....trees. Section 6 explains how we evaluated the gathercast protocol. The paper concludes with a description of possible extensions of this work. 2 Related Work There are mechanisms that require network participation for increasing the goodput. For example, many reliable multicast schemes [9, 10, 11, 12] propose suppressing NACKs to solve the problem of NACK implosion. In such schemes, if a node within the network receives identical NACKs from multiple nodes, it forwards just one NACK to the source. The Cu SeeMe project at the Cornell University [13] advocates the concept of reflectors which are ....

L. wei H. Lehman, S. J. Garland, and D. L. Tennenhouse, "Active reliable multicast," in Proceedings of the INFOCOM, 1998.

....The ACTIVE IP Option [4] embeds small Tcl programs in the option fields of IPv4 packets. The ANTS system [5] supports rapid prototyping and applicationspecific in network processing through a capsule based system written in Java. Several sample applications have been developed using ANTS [5] [6], 7] The development of PAN drew heavily on many of the ideas developed in the work on ANTS. Unlike ANTS, however, PAN is designed for performance so that it can be used for real applications. PAN introduces the ideas of software segments and panStreams, has a more flexible system for assembling ....

Li-Wei H. Lehman, Stephen J. Garland, and David L. Tennenhouse, "Active Reliable Multicast," in IEEE Infocom 1998, San Francisco, California, April 1998.

No context found.

L. wei H. Lehman, S. J. Garland, and D. L. Tennenhouse. Active reliable multicast. In Proceedings of IEEE INFOCOM '98, 1998.

No context found.

L. Wei, H. Lehman, S. J. Garland, and D. L. Tennenhouse. Active reliable multicast. In IEEE INFOCOM'98.

No context found.

L. Wei, H. Lehman, S. J. Garland, and D. L. Tennenhouse. Active reliable multicast. In IEEE INFOCOM '98, March 1998.

No context found.

L. Wei, H. Lehman, S. J. Garland, and D. L. Tennenhouse. Active reliable multicast. In IEEE INFOCOM'98.

No context found.

L. Wei, H. Lehman, S. J. Garland, and D. L. Tennenhouse. Active reliable multicast. In IEEE INFOCOM'98.

No context found.

Li wei H. Lehman, Stephen J. Garland, and David L. Tennenhouse. Active reliable multicast. In IEEE INFOCOM'98, 1998. 7

No context found.

Li-Wei Lehman, Stephen Garland, and David Tennenhouse, "Active Reliable Multicast," in Proc. IEEE Infocom '98, 1998, pp. 581--589.

First 50 documents

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