ANDREWS, M., AND ZHANG, L. 2000. The effects of temporary sessions on network performance. In Proceedings of the 11th Annual ACM--SIAM Symposium on Discrete Algorithms (San Francisco, Calif., Jan. 9 --11). ACM, New York, pp. 448 -- 457.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....Systems General Terms Algorithms, Performance, Theory Keywords Adversarial queuing theory 1. INTRODUCTION Recent years have seen a growing amount of work being concentrated on analyzing packet switching networks under non probabilistic scenarios rather than under probabilistic assumptions [6, 2, 4, 12, 10, 11, 1, 3]. Much of this work makes use of the model of adversarial queuing theory proposed by Borodin et al. 6] The basic model can be briefly described as follows. Time proceeds in discrete steps. In each step, packets are injected into the system with their routes. Each packet traverses its ....

M. Andrews, and L. Zhang, The Effect of Temporary Sessions on Network Performance. In Proc. of the 11th SODA, 2000.

....when r: 1. During the course of the proof we obtained the following bound on the total number of packets in the network where B0 is the initial number of packets in the network and Pm is the maximal length Unfortunately, the bound is exponential in the network parameters. It is shown in [4] that both NTO and FTG lead to exponentially large queue sizes in certain networks. It is also shown that no bound better than 2 m IVI,IEI is possible for a whole class of distributed deterministic policies including NTG, FTG. Note that, unlike r 1 case, stability under r: 1 condition does not ....

M. Andrews and L. Zhang. The effects of temporary sessions on network performance. Proc. 11th A CM-SIAM Symposium on Discrete Algorithms, 2000.

....that this fails to hold when the hot potato constraint is present and the ratio can take a variety of values depending on the structure of the graph. The third line of research is the study adversarial queuing networks initiated by Borodin et al. 8] and followed by several other works [3] 2] [4] [12] 1] 11] 10] Simple stable scheduling policies were constructed in [3] and [10] which achieve stability as long as the arrival rate r 1. We will show in this paper that this fails to hold in most of the networks if hot potato constraint on the schedule needs to be satisfied. 1.3 Our ....

M. Andrews and L. Zhang. The effects of temporary sessions on network performance. Proc. 11th A CM-SIAM Symposium on Discrete Algorithms, 2000.

No context found.

ANDREWS, M., AND ZHANG, L. 2000. The effects of temporary sessions on network performance. In Proceedings of the 11th Annual ACM--SIAM Symposium on Discrete Algorithms (San Francisco, Calif., Jan. 9 --11). ACM, New York, pp. 448 -- 457.

....to determine whether such a protocol exists. We note that the randomized protocol of Section 4.2 can be converted into a deterministic, centralized protocol with polynomially bounded queues; thus, the emphasis is on finding a protocol that is both deterministic and distributed. In recent work [3], Andrews and Zhang show that LIS can produce delays of Omega Gamma e ) More generally they show that for any deterministic protocol that selects the packet to advance independently of the packet routes, the delay bound cannot be better than O(e ) Hence a determinisitic, distributed ....

M. Andrews and L. Zhang. The effects of temporary sessions on network performance. In Proc. 11th ACM-SIAM SODA, 2000.

....were injected to change over time. This model was later considered in [2] and a number of simple, deterministic protocols were analyzed. Some of these protocols were shown to produce finite delays, however all the delay bounds were exponential in the length of the longest path. The papers [2] and [5] show that all these protocols can in fact produce delays that are exponential in the length of the longest path. In [2] a randomized protocol was presented for which the delay bound was O( dmax log m) i.e. the delay bound was polynomial in the network size and longest path length. However, ....

M. Andrews and L. Zhang. The effects of temporary sessions on network performance. In Proceedings of the 11th Annual ACM-SIAM Symposium on Discrete Algorithms, pages 448 -- 457, San Francisco, CA, January 2000.

....with N D= w Gamma 2 D) 1, the number of (N Gamma 1; N) packets is more than D and hence the delay bound will be violated. Therefore D w=2 = Theta(w= We also note that the delay bound of LIN is exponential in the number of switches on the session paths. This also seems unavoidable since [17] shows that for temporary sessions in networks of output queued switches most deterministic, distributed protocols are unable to produce subexponential delay bounds. This contrasts with the situation for permanent sessions where the classical delay bounds for protocols such as Weighted Fair ....

M. Andrews and L. Zhang, "The effects of temporary sessions on network performance," in Proceedings of the 11th Annual ACM-SIAM Symposium on Discrete Algorithms, San Francisco, CA, January 2000, pp. 448 -- 457.

No context found.

ANDREWS, M., AND ZHANG, L. 2000. The effects of temporary sessions on network performance. In Proceedings of the 11th Annual ACM--SIAM Symposium on Discrete Algorithms (San Francisco, Calif., Jan. 9 --11). ACM, New York, pp. 448 -- 457.

No context found.

M. Andrews, and L. Zhang, The Effect of Temporary Sessions on Network Performance. In Proc. of the 11th SODA, pp. 448-457, 2000.

No context found.

M. Andrews, and L. Zhang, The Effect of Temporary Sessions on Network Performance. In Proc. of the 11th SODA, 2000.

No context found.

Matthew Andrews and Lisa Zhang, "The effects of temporary sessions on network performance," in Proceedings of the Symposium on Disscrete Algorithms (SODA), 2000, To appear.

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