Kar, K., Kodialam, M. and Lakshman, T.V., Minimum Interference Routing of Bandwidth Guaranteed Tunnels with MPLS Tra#c Engineering Applications, IEEE Journal of Selected Areas in Communications, vol. 18, no. 12, pp. 2566-2579, 2000.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....be fundamentally difficult in packet based networks like the Internet [21] Consequently, there has been considerable interest in traffic engineering using explicitly routed paths. While explicit routes can be chosen in many ways, most recent solutions, such as Minimum Interference Routing (MIRA) [7] and Profile based Routing (PBR) 17] come in the context of MPLS based routing. The challenge with these approaches is the need to keep state at the per flow level, which limits the scalability of the solution. The other challenge is the requirement of having extra information in the form of ....

....producing a distribution with finite variance to have smaller confidence intervals [16] The total load is defined as = P K i=1 i = i , where K is number of classes. Given i , b i , and i , B i is computed by the following equation: Widest Shortest Path [6] Minimum Interference Routing [7] Profile based Routing [17] Dynamic per packet routing Solves Guaranteed bandwidth Guaranteed bandwidth Guaranteed bandwidth Best Effort Routing Strategy Source Source Source Hop by hop Type Link State Link State Link State Link State or Distance Vector Time Complexity O(ElogV) 3] O(nVE 2 ) ....

[Article contains additional citation context not shown here]

K. Kar, M. Kodialam, and T. Lakshman. "Minimum Interference Routing of Bandwidth Guaranteed Tunnels with MPLS Traffic Engineering Applications". In Proceedings of IEEE INFOCOM, Tel-Aviv, Israel, March 2000.

.... engineering control polA has been extended and generalized to serve as the controlpntr for di#erenttypr of switched transpan networks, ranging frompomAVE switched networks and time division multipn xingcapAVR interface (TDM) technologies, to automatically switchedop0EE ltransp rt networks [3,20,21]. This generic MPLS based controlpntr technology is Computer pApGA ly being standardized by the IETF within the concep of generalized MPLS (GMPLS) The advent of MPLS along with the standardization issues surrounding it has psA oked a significant organizational realignment within the IETF, ....

....heuristics. Among the early work in the area of o#ine MPLS constraint based routing is the contribution by Fahim [19] which utilized a centralized global opal ization algorithm. Aheuristic onlinepli selection method, referred to as minimum interference routing algorithm (MIRA) isp22QNN ed in [20]. This algorithm, which requires ap20VV knowledge of ingress egressp irs,attempN to defer loading of certain so called critical links . The critical link is defined as those links whose congestion will cause blocking of future LSPsetup requests between more than one ingress egress p ir. ....

K. Kar, M. Kodialam, T.V. Lakshman, Minimum interference routing of bandwidth guaranteed tunnels with MPLS tra#c engineeringapineeringAE IEEE Journal on Selected Areas in Communications 18 (12) (2000).

....they act only when problems start to appear. 2.1 Preventive schemes Constraint based routing schemes in MPLS networks have their roots in the well known Quality of Service routing problems in IP networks. One of the most cited schemes, called MIRA (Minimum Interference Rout ing Algorithm) [9], is based on an heuristic dynamic online path selection al gorithm. The key idea is to exploit the a priori knowledge of ingress egress pairs to avoid routing over links that could interfere with potential future paths set up. These critical links are identified by MIRA as links that, if ....

....LSPa can cross its shortest path through the network. 5.2 Path set up rejection ratio This section describes the set of experiments used to compare the performance of DYLBA in term of path set up rejection ratio with MHA and MIRA. These experiments are carried out by using network topology of [9], see Figure 5. The links are all bidirectional with a capacity of 120 units (thin lines) and 480 units (thick lines) These values are taken to model the capacity ratio of OC 12 and OC 48 links. Path requests are limited only to the ingress and egress router pairs (S,D) S2,D2) Sa,Da) and All ....

K. Kar, M. Kodialam, and T.V. Lakshman. Minimum Interference Routing of Bandwidth Guaranteed Tunnels with MPLS Traffic Engineering Appli- cations. IEEE Journal on Selected Areas in Communications, 18(12):2566 -2579, December 2000.

....maximum delay, administrative policies) The second mechanism, traffic splitting, balances the network load through optimal partitioning of traffic to parallel LSPs between pairs of ingress and egress nodes. One of the most cited CBR schemes, called MIRA (Minimum Interference Routing Algorithm) [9], is based on an heuristic dynamic online path selection algorithm. The key idea is to exploit the a priori knowledge of ingress egress pairs to avoid routing over links that could interfere with potential future paths set up. These critical links are identified by MIRA as links that, if ....

....a constant computational time. The value of depends on the average bandwidth of the LSPs in the network and the link capacities. Therefore the complexity of our algorithm is not more than W c . V. SIMULATION RESULTS The simulations are carried out by using the network topology of [9], see Figure 3. The links are all bidirectional with a capacity of 120 units (thin lines) and 480 units (thick lines) These values are taken to model the capacity ratio of OC 12 and OC 48 links. Path requests are limited only to the ingress and egress router pairs ( X , and X , ....

[Article contains additional citation context not shown here]

K. Kar, M. Kodialam, and T.V. Lakshman. Minimum Interference Routing of Bandwidth Guaranteed Tunnels with MPLS Traffic Engineering Applications. IEEE Journal on Selected Areas in Communications, 18(12):2566 --2579, December 2000.

....from ingress node s to egress node d. We also refer informally to paths by their node sequences ordered from s to d, e.g. s u We also associate with a trunk its effective bandwidth which we assume to account for the trunk s QoS requirements (e.g. constraints on packet delay, jitter, or loss) [2] and to represent the amount of bandwidth that is reserved for the trunk if it is routed. Although there are different definitions of effective bandwidth [3 9] in general, it is a value intermediate between the mean and peak flow rates of a connection or Costp Cost. 3) as the corresponding ....

IC Kar, M. Kodialam, T. V. Lakshman, "Minimum interference routing of bandwidth guaranteed runnels with IVlPLS traffic engineering applications," IEEE Journal on Selected Areas in Communications, vol. 18, no. 12, December 2000.

....can be defined (and solved with heuristics) where LSPs are set up to meet constraints in terms of both the traffic demands and the resources available. OSPF traffic engineering extensions [20] can be used to collect the information needed to setup this problem. Variations of the max flow technique [19] or adaptive traffic splitting [9] can be used to map traffic onto the LSPs. In the inter domain area, the IRTF is considering requirements documents for a future inter domain protocol, and traffic engineering figures in both the key proposals [8, 17] as an important problem. Inter domain TE work ....

K. Kar, M. Kodialam, T. V. Lakshman, "Minimum Interference Routing of Bandwidth Guaranteed Tunnels with MPLS Traffic Engineering Applications," IEEE JSAC, Vol. 18, No. 12, December 2000.

....layout design and flow assignment. It would be e#cient to run o# line algorithms if we have a priori knowledge about tra#c demands and patterns. But such assumption is not valid in practice. Some online algorithms have been proposed to address LSPs layouts and flow assignments for unicast tra#c [7, 9, 10]. However, to the best of our knowledge, no algorithms have been proposed yet for tra#c engineering of multicast flows in MPLS domains. 1.2 Di#culties in Supporting IP Multicast in MPLS Domains As MPLS is standardized by IETF and is expected to be implemented in the near future, it is inevitable ....

K. Kar, M. Kodialam, and T.V.Lakshman, Minimum Interference Routing of Bandwidth Guaranteed Tunnels with MPLS Tra#c Engineering Applications , IEEE Journal on selected areas in communications, pp. 2566-2579, Vol18, No.12, December 2000.

....be ingress egress routers, between which label switched paths (LSPs) can be set up. We assume that the ingress egress pairs are known, and that this information is quasi static, i.e. that it changes very infrequently. An example is shown in Figure 2, which is borrowed from Kar Kodialam Lakshman [KKL00] We call this network the KL1 graph, and it is one of the several networks we used for our simulations. A request for an LSP setup is defined by a quadruple (id, s i , d i , b i ) where id is the request ID, s i is the ingress (source) router, d i is the egress (destination) router, and ....

....we used for our simulations. A request for an LSP setup is defined by a quadruple (id, s i , d i , b i ) where id is the request ID, s i is the ingress (source) router, d i is the egress (destination) router, and This figure actually reflects the graph used for the MIRA measurements in [KKL00] although, according to one of the authors, the figure actually printed in their paper does not correctly reflect all high bandwidth links. D3 S2 D4 D2 S4 D1 1 3 2 5 12 14 13 9 15 8 4 7 6 11 Figure 2: An example network, showing ingress egress nodes. b i is the bandwidth ....

[Article contains additional citation context not shown here]

K. Kar, M. Kodialam, and T.V. Lakshman. Minimum interference routing of bandwidth guaranteed tunnels with MPLS tra#c engineering applications. IEEE Journal on Selected Areas in Communications, 18(12):2566--2579, December 2000.

....are known Knowledge of ingress egress pairs can be used along with QoS routing (Section 3) to improve utilization and long term performance. For bandwidth constrained on demand routing where there is no knowledge of future requests, and splitting of individual requests is not allowed, Kar et al. [7] suggest picking feasible paths that interfere least with future requests. Figure 4 shows an example. When S3 wants 10 6 1 23 4 5 7 8 9 11 S1 S2 S3 D1 D2 D3 Figure 4. From S3 to D3, 1 2 3 4 5 is the minimum interference path. to communicate with D3, either P 1 (1,7,8,5) or P 2 ....

....upper bound on the total amount of bandwidth that can be routed between an ingress egress pair. A linear programming formulation cannot be used to maximize the sum of maximum flows between other ingressegress pairs because of splitability restrictions. Therefore, the problem is NP hard. Kar et al. [7] propose a heuristic which is called Minimum Interference Routing Algorithm (MIRA) The idea is to assign link weights based on their criticality. A link is critical if it belongs to a mincut for an ingress egress pair. In other words, routing over a critical link would decrease the maxflow value ....

[Article contains additional citation context not shown here]

K. Kar, M. Kodialam, and T. Lakshman. "Minimum Interference Routing of Bandwidth Guaranteed Tunnels with MPLS Traffic Engineering Applications". In Proceedings of IEEE INFOCOM, Tel-Aviv, Israel, March 2000.

....be fundamentally difficult in packet based networks like the Internet [21] Consequently, there has been considerable interest in traffic engineering using explicitly routed paths. While explicit routes can be chosen in many ways, most recent solutions, such as Minimum Interference Routing (MIRA) [7] and Profile based Routing (PBR) 17] come in the context of MPLS based routing. The challenge with these approaches is the need to keep state at the per flow level, which limits the scalability of the solution. The other challenge is the requirement of having extra information in the form of ....

....capture the heavy tail nature of connection durations while still producing a distribution with finite variance to have smaller confidence intervals [16] The total load is defined as # ### # = # , where K is number of classes. Given 3 Widest Shortest Path [6] Minimum Interference Routing [7] Profile based Routing [17] Dynamic per packet routing Solves Guaranteed bandwidth Guaranteed bandwidth Guaranteed bandwidth Best Effort Routing Strategy Source Source Source Hop by hop Type Link State Link State Link State Link State or Distance Vector Time Complexity O(ElogV) 3] O(nVE ) ....

[Article contains additional citation context not shown here]

K. Kar, M. Kodialam, and T. Lakshman. "Minimum Interference Routing of Bandwidth Guaranteed Tunnels with MPLS Traffic Engineering Applications". In Proceedings of IEEE INFOCOM, Tel-Aviv, Israel, March 2000.

....to be ingress egress routers, between which label switched paths (LSPs) can be set up. We assume that the ingress egress pairs are known, and that this information is quasi static, i.e. that it changes very infrequently. An example is shown in Figure 2, which is borrowed from Kar Kodialam Lakshman [14]. We call this network the KL1 graph, and it is one of the several networks we used for our simulations. A request for an LSP setup is defined by a quadruple (id, s i , d i , b i ) where id is the request ID, s i is the ingress (source) router, d i is the egress (destination) router, and b i ....

....or reject it. The admission control feature allows our algorithm to reject a request even if there is a feasible path this may happen if the algorithm determines that accepting this request may create a significant bot1 This figure actually reflects the graph used for the MIRA measurements in [14], although, according to one of the authors, the figure that was printed in the paper does not correctly reflect all high bandwidth links. Fig. 2. An example network, showing ingress egress nodes. tleneck for future requests (based on its knowledge of the ingress egress pairs and their tra#c ....

[Article contains additional citation context not shown here]

K. Kar, M. Kodialam, T. Lakshman, Minimum interference routing of bandwidth guaranteed tunnels with MPLS tra#c engineering applications, IEEE Journal on Selected Areas in Communications 18 (12) (2000) 2566--2579. 29

....performance by improved path selection. For some alternative approaches to this problem in the context of optical networks, see [3] and the references therein. For dynamic routing of non restorable connections, a recently proposed algorithm that performs very well is minimum interference routing [9]. Here the idea is that a newly routed connection s path must not interfere too much with paths that might be critical to satisfy future demands (as explained in more detail later) In this paper, we develop two new algorithms for routing restorable connections where the objective is to improve ....

....new algorithms for routing restorable connections where the objective is to improve performance by routing using the minimum interference criteria. Note that even though we use the minimum interference criteria, the developed algorithms are new and are not simple extensions of the algorithms in [9]. However, as with the algorithms in [9] the only information needed for routing is the link residual bandwidths and knowledge of network ingress egress pairs is exploited to achieve performance improvements. No assumptions are made regarding knowledge of future arrivals and connections arrive ....

[Article contains additional citation context not shown here]

K. Kar, M. Kodialam, T. V. Lakshman, "Minimum Interference Routing of Bandwidth Guaranteed Tunnels with MPLS Traffic Engineering Applications ", IEEE Journal on Selected Areas in Communications, Vol. 18, No. 12, December 2000.

....paper is only on the no sharing case. Since we cannot achieve efficiency by sharing, our objective here is to improve performance by improved path selection. For dynamic routing of non restorable connections, a recently proposed algorithm that performs very well is minimum interference routing [8]. Here the idea is that a newly routed connection s path must not interfere too much with paths that might be critical to satisfy future demands (as explained in more detail later) In this paper, we develop two new algorithms for routing restorable connections where the objective is to improve ....

....new algorithms for routing restorable connections where the objective is to improve performance by routing using the minimum interference criteria. Note that even though we use the minimum interference criteria, the developed algorithms are new and are not simple extensions of the algorithms in [8]. However, as with the algorithms in [8] the only information needed for routing is the link residual bandwidths and knowledge of network ingress egress pairs is exploited to achieve performance improvements. No assumptions are made regarding knowledge of future arrivals and connections arrive ....

[Article contains additional citation context not shown here]

K. Kar, M. Kodialam, T. V. Lakshman, "Minimum Interference Routing of Bandwidth Guaranteed Tunnels with MPLS Traffic Engineering Applications", IEEE Journal on Selected Areas in Communications, Vol. 18, No. 12, December 2000.

No context found.

Kar, K., Kodialam, M. and Lakshman, T.V., Minimum Interference Routing of Bandwidth Guaranteed Tunnels with MPLS Tra#c Engineering Applications, IEEE Journal of Selected Areas in Communications, vol. 18, no. 12, pp. 2566-2579, 2000.

No context found.

K. Kar, M. Kodialam, and T. V. Lakshman, Minimum interference routing of bandwidth guaranteed tunnels with MPLS traffic engineering applications, IEEE JSAC, Vol. 18(12), pp. 2566-2579, Dec. 2000.

No context found.

Kar, K., Kodialam, M. and Lakshman, T.V., Minimum Interference Routing of Bandwidth Guaranteed Tunnels with MPLS Tra#c Engineering Applications, IEEE Journal of Selected Areas in Communications, vol. 18, no. 12, pp. 2566-2579, 2000.

No context found.

K. Kar, M. Kodialam, and T. V. Lakshman, Minimum interference routing of bandwidth guaranteed tunnels with MPLS traffic engineering applications, IEEE JSAC, Vol. 18(12), pp. 2566-2579, Dec. 2000.

No context found.

Koushik Kar, Murali Kodialam, and T. V. Lakshman. Minimum interference routing of bandwidth guaranteed tunnels with MPLS tra#c engineering applications. IEEE J. on Sel. Areas in Commun., 18(12):2566--2579, December 2000.

No context found.

K. Kar, M. Kodialam, and T.V. Lakshman. Minimum Interference Routing of Bandwidth Guaranteed Tunnels with MPLS Traffic Engineering Applications. In INFOCOM (2), pages 884--893, 2000.

No context found.

K. Kar, M. Kodialam, and T. V. Lakshman, Minimum interference routing of bandwidth guaranteed tunnels with MPLS traffic engineering applications, IEEE Journal on Selected Areas in Communications: Quality of Service in the Internet, Vol. 18, No. 12, pp. 921--940, Dec. 2000.

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