N. F. Maxemchuk. Comparison of deflection and store and forward techniuques in the Manhattan street and shu#e exchange networks. In Proc. IEEE INFOCOM, pages 800--809, 1989.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....is treated like a hot potato . Hot potato routing algorithms have been observed to work well in practice [4] and have been used in parallel machines such as the HEP multiprocessor [25] the Connection machine [13] and the Caltech Mosaic C [24] as well as in high speed communication networks [19]. Hot potato routing algorithms are well suited for optical networks since it is di#cult to bu#er optical messages [1, 26] 1 L 0 1 2 3 4 5 6 0 1 2 3 Butterfly Mesh L 1 . 0 2 Leveled Network Figure 1: Leveled networks We study hot potato routing algorithms for the wide class ....

N. F. Maxemchuk. Comparison of deflection and store and forward techniuques in the Manhattan street and shu#e exchange networks. In Proc. IEEE INFOCOM, pages 800--809, 1989.

....transmitter arrays, two receiver arrays and a 2 Theta 2 switch for routing purposes. Information is retained in the optical domain from source to destination except for the packet headers decoded at each intermediate node. The approach has no buffering at the switches and uses deflection routing [10 12]. For the rest of this paper, the term SHUFFLENET refers to this architecture. WDM networks are typically built using at least one wavelength tunable transceiver component. One of the important factors determining the cost and feasibility of a network architecture is the transceiver component ....

N. F. Maxemchuk, "Comparison of deflection and store-and-forward techniques in the Manhattan street and shuffle-exchange networks," in Proc. IEEE INFOCOM'89, vol. 3, (Ottawa, Canada), pp. 800--809, Apr. 1989.

....in worm LANs is deflection routing. The scheme works as follows: if the worm arriving at the switch finds its preferred output (as defined by a routing table or by a source routing header) busy, it is deflected to any other output which happens to be free. The first deflection routing studies [5, 15, 16] were aimed at synchronous, slotted networks. Later, slotted and unslotted deflection routing for packets of fixed size was studied in [6] Recently, deflection in asynchronous networks was addressed in [18, 19] Deflection routing in unslotted, asynchronous 4 networks with wormhole routing is ....

N. Maxemchuk. Comparison of Deflection and Store-and-Forward Techniques in the Manhattan Street and Shuffle-Exchange Networks. In Proceedings of INFOCOM 89, volume 3, pages 800--809, April 1989.

....at a switch on one of its input ports, at least one output port must be idle. Thus, there is always a way to relay the packet, although it may not always be the best possible way. No packet is ever blocked at a switch. This routing mechanism resembles the so called deflection routing introduced in [23, 24] for Manhattan Street and Shuffle Exchange networks. In our case, the operation of the network is completely asynchronous and no packets are ever buffered at the switches. If k h 0, i.e. there are some hosts connected to the switch, the switch is expected to recognize packets addressed to ....

....cost effective in wide area applications, not only due to the geographic constraints, but also from the viewpoint of traffic characterization. 26 6 Relation to previous work and suggestions for further research Numerous issues related to deflection networks have been discussed in the literature [1, 3, 5, 10, 11, 16, 17, 18, 23, 24, 25]. Most of the models considered there are based on two simple and regular network architectures proposed by Maxemchuk in [23, 24] namely the Manhattan Street and Shuffle Exchange networks [5, 10, 11, 16, 18, 25] Notably, Greenberg and Hajek [17] investigate the performance of synchronous ....

[Article contains additional citation context not shown here]

N. Maxemchuk. Comparison of deflection and store-and-forward techniques in Manhattan-street network and shuffle-exchange networks. In Proceedings of IEEE INFOCOM'89, pages 800--809, 1989.

....links is the same it is always possible to forward a packet from an incoming link to some outgoing link no buffers are required. Most of the results obtained so far are for mesh and torus and therefore more applicable to multiprocessors one site computers than to communication networks. In [Ma89] it is proposed to apply deflection routing to communication network with regular structure (the Manhattan street network) In a recent work [NOR95] it is suggested to use scheduled hot potatoes approach for a restricted type of high speed networks. In [NOR95] it is assumed that packets do not ....

N.F. Maxemchuk, "Comparison of deflection and store and forward techniques in the Manhattan street and shuffle exchange networks", Proc. IEEE INFOCOM, pp. 800-809, 1989.

....360 MIPS 90 Mhz and 64 MB of memory for each processor. 5.2 Applications 5.2.1 Shuffle Exchange Network We simulated three shuffle exchange networks a 100x100, a 165x165, and a 200x200 SXN. The networks had a total of 36,951, 100,102 and 146,672 LPs respectively. The Shuffle Exchange Network (Maxemchuk 1989; Robertazzi 1993) is a cylindrical multi stage network with an in and out degree of two. For a stress test, we modified the original shuffle exchange network by interconnecting nodes in the first and last column and substituting one of input buffers with a local source and one of output buffers ....

Maxemchuk, N. 1989. "Comparison of Deflection and Storeand -Forward Techniques in the Manhattan Street and ShuffleExchange Networks." In Proceedings of IEEE Infocom., 800-- 809.

....routing algorithm there is no buffering of packets at nodes; i.e. each packet must traverse a link at every step until it reaches its destination. This approach, introduced some 35 years ago by Baran [4] has been observed in a number of experiments to perform exceptionally well in practice [1, 11, 19, 20, 24, 25], and has been used in parallel machines such as the HEP multiprocessor [31] the Connection Machine [22] and the Caltech mosaic C [30] The elimination of buffering queues used in store and forward algorithms has the advantage of potentially faster switching, which is particularly important for ....

N. F. Maxemchuk, Comparison of deflection and store-and-forward techniques in the Manhattan street and shuffle-exchange networks. In Proc. INFOCOM, pp. 800--809, IEEE, 1989. 20

....are many differing methods of deciding when to deroute packets and choosing which packet to deroute. However, two basic classes of non minimal adaptive routers can be distinguished based on their buffering characteristics: deflection routers and queueing routers. Deflection routers [Smith 81, Maxemchuk 89, Fang Szymanski 91, Smitley 89] also known as hot potato or desperation routers, use a synchronous approach and a time step long enough to transmit an entire packet. At each step the incoming packets are paired with outgoing channels and are transmitted in the next step. Since deflection ....

N. F. Maxemchuk. Comparison of deflection and store-and-forward techniques in the manhattan street and shuffle-exchange networks. In Proceedings of IEEE INFOCOM '89, pages 800--809. IEEE, 1989.

....and Computer Science, Haifa University, Haifa, Israel 31905. E mail: ilan mathcs.haifa.ac.il Department of Computer Science, Technion, Haifa, Israel 32000. E mail: assaf cs.technion.ac. il 1 Introduction In this work we consider a routing mode known as hot potato or deflection routing [AS91, GG86, GH92, Haj91, Max89, Szy90, ZA91, NS89]. The important characteristic of algorithms which assume this mode is that they use no buffer space for storing delayed packets. Each packet, unless it has already reached its destination, must leave the processor at the step following its arrival. Packets may reach a processor from all its ....

....to its preferred direction. Much work has recently focused on exact analysis and design of hot potato routing algorithms. One reason is that variants of hot potato routing are used by recent parallel machines such as the HEP multiprocessor [Smi81] and by high speed communication networks [Max89]. In particular, hot potato routing is very important in fine grain massively parallel computers, such as the Caltech Mosaic C [Sei92] For such machines, even the inclusion of a small sized storage buffer at each processor causes a substantial increase in the cost of the machine. Another domain ....

N.F. Maxemchuk. Comparison of deflection and store and forward techniques in the manhattan street and shuffle exchange networks. In IEEE INFOCOM, pages 800--809, 1989.

....are many different methods of deciding when and which packet to deroute. However, based on their buffering characteristics, the three basic classes of non minimal adaptive routers are: deflection routers, queuing routers, and wormhole routers. C.1.a Deflection Routing. Deflection routers [20] [21], 22] 23] also known as hot potato or desperation routers, use a synchronous approach and a time step long enough to transmit an entire packet. At each step the incoming packets are paired with outgoing channels and are transmitted in the next step. Since deflection routers use a synchronous ....

N. F. Maxemchuk, "Comparison of deflection and store-andforward techniques in the manhattan street and shuffle-exchange networks," in Proceedings of IEEE INFOCOM '89, 1989, pp. 800--809. 12 IEEE TRANSACTIONS ON COMPUTERS, VOL. XX, NO. YY, MONTH 199X

....topology for gigabit networks, since it can cover a large geographical area with small total fiber length. Also, because of its regularity and symmetry properties, the MS network has been analyzed extensively in the literature for datagram deflection schemes (see, for example, 4] 6] 8] 11] [15], 16] 27] We believe that the results obtained are characteristic of the performance of the VCD protocol for other topologies that offer, as the MS network does, a large number of alternative paths between any source destination pair of nodes. The dimensional wraparound mesh consists of ....

N. F. Maxemchuk, "Comparison of deflection and store-and-forward techniques in the Manhattan Street and Shuffle-Exchange networks," in Proc. IEEE INFOCOM'89, vol. 3, pp. 800--809.

....1 packets in dn O(log 2 n) time steps on the d dimensional n d node torus and in 2dn O(log 2 n) time steps on the d dimensional n d node mesh, with high probability. This research was partially supported by the NEC Research Institute. 1 Introduction A number of researchers [1, 4, 3, 2, 5, 6, 7, 8, 10, 11, 14, 12] have suggested algorithms for routing packets in a network with the property that on each step, each node in the network sends all of the packets it received on the previous step along one of its outgoing edges (with at most one packet leaving per edge) Such schemes are generally referred to as ....

....of which edge to send the incoming packets out is a simple one, then the resulting algorithm is easy to implement with minimal time between routing steps. Variants of hot potato routing are used in the Connection Machine [9] the HEP multiprocessor [16] and in high speed communications networks [12]. While the apparent advantages of hot potato algorithms have been borne out by numerous simulation studies [1, 6, 7, 11, 12, 15] exact analysis of their behavior (i.e. without any independence assumptions) has proven to be difficult. Hajek [8] showed that for a natural algorithm on the N node ....

[Article contains additional citation context not shown here]

Maxemchuk, N., "Comparison of deflection and store and forward techniques in the Manahattan street and shuffle-exchange networks", IEEE INFOCOM, 1989, 800--809.

....: 75 Chapter 1 Introduction 1.1 Background This work studies packet routing in synchronous network of processors, in which any communication link can carry at most one packet at each time step. We consider a form of packet routing known as hot potato routing or deflection routing [1, 12, 13, 15, 21, 27, 28]. The important characteristic of this form is that it requires no buffer space for storing delayed packets. Each packet must leave the processor at the step following its arrival, unless it has arrived to its destination. Packets arriving to a processor from its neighbors have to be redirected to ....

....as hot potato routing trades off memory resources to communication links. The desirability of such tradeoff depends on the available resources and technology. Parallel machines such as the Connection Machine [16] and the HEP multiprocessor [26] as well as high speed communication networks [21] use deflection routing in various forms. Deflection routing is also highly desirable in the domain of optical networks [1, 13, 27, 28] In such networks, the storing of packets require conversions between optical form and electronic one. Currently, these conversions are much slower than optical ....

N.F. Maxemchuk. "Comparison of deflection and store and forward techniques in the manhattan street and shuffle exchange networks." IEEE INFOCOM, pp. 800-8009, 1989.

....space. Packets that at the moment of arrival cannot be relayed via their preferred routes (because those routes are busy) are deflected, i.e. relayed via suboptimal routes. This concept, traditionally illustrated by the Manhattan street networks (MSN) introduced and analysed by Maxemchuk in [8, 9, 10, 11, 12], works under the assumption that the amount of buffer space available at a switch is sufficient to determine the route preference of an incoming packet before the packet is relayed via one of the outgoing links. No buffer space is required for storing packets that cannot be immediately relayed ....

N. Maxemchuk. Comparison of deflection and store-and-forward techniques in Manhattanstreet network and shuffle-exchange networks. In Proceedings of IEEE INFOCOM'89, pages 800--809, 1989.

....switch. Packets that at the moment of arrival cannot be relayed via their preferred routes (because those routes are busy) are deflected, i.e. relayed via suboptimal routes. This concept, traditionally illustrated by the Manhattan street networks (MSN) introduced and analyzed by Maxemchuk in [11, 12, 13, 14, 15], works under the assumption that the amount of buffer space available at a switch is sufficient to determine the route preference of an incoming packet before the packet is relayed via one of the outgoing links. No buffer space is required for storing packets that cannot be immediately relayed ....

N. Maxemchuk. Comparison of deflection and store-and-forward techniques in Manhattan-street network and shuffle-exchange networks. In Proceedings of IEEE INFOCOM'89, pages 800--809, 1989.

No context found.

N. F. Maxemchuk. Comparison of deflection and store and forward techniuques in the Manhattan street and shu#e exchange networks. In Proc. IEEE INFOCOM, pages 800--809, 1989.

No context found.

N. F. Maxemchuk. Comparison of deflection and store and forward techniuques in the Manhattan street and shu#e exchange networks. In Proc. IEEE INFOCOM, pages 800--809, 1989.

No context found.

N. F. Maxemchuk. Comparison of deflection and store and forward techniuques in the Manhattan street and shu#e exchange networks. In Proc. IEEE INFOCOM, pages 800--809, 1989.

No context found.

N. F. Maxemchuk. Comparison of deflection and store-and-forward techniques in the manhattan street and shuffle-exchange networks. In INFOCOM, Eighth Annual Joint Conference of the IEEE Computer and Communications Societies, volume 3, 1989.

No context found.

N.F. Maxemchuk, Comparison of deflection and store and forward techniques in the Manhattan street and shuffle exchange networks, In IEEE INFOCOM, (1989) 800--809.

No context found.

N. F. Maxemchuk, "Comparison of deflection and store and forward techniques in the Manhattan street network and shuffle exchange networks, " in Proc. IEEE INFOCOM '89, pp. 800--809.

No context found.

N. Maxemchuk. Comparison of deflection and storeand -forward techniques in Manhattan-street network and shuffle-exchange networks. In Proceedings of IEEE INFOCOM'89, pages 800--809, 1989.

No context found.

N. F. Maxemchuk. Comparison of deflection and store and forward techniuques in the Manhattan street and shu#e exchange networks. In Proc. IEEE INFOCOM, pages 800--809, 1989.

No context found.

N. F. Maxemchuk, "Comparison of deflection and store-and-forward techniques in the Manhattan Street and ShuffleExchange networks," in Proceedings of IEEE Infocom, pp. 800--809, 1989.

No context found.

N. F. Maxemchuk. Comparison of deflection and store and forward techniuques in the Manhattan street and shu#e exchange networks. In Proc. IEEE INFOCOM, pages 800--809, 1989.

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