This document proposes the Minimum Interference Routing Algorithm (MIRA) to optimize routing in an MPLS network. MIRA aims to increase speedy traffic availability while minimizing resource usage. It works by identifying critical links, computing weights for paths, and finding the minimum interference route using Dijkstra's algorithm. The authors model an MPLS network, apply three versions of MIRA, and analyze routing interference and network performance improvements compared to shortest path routing. They conclude that MIRA reduces traffic interference and congestion, optimizing network resources and improving performance metrics.
Routing Management with MIRA and enhancement (IMECS2010)
1. Analysis and Simulation Routing Management MIRA (Minimum Interference Routing Algorithm) for Speedy traffic on MPLS Network By : 1. RendyMunadi (rnd@ittelkom.ac.id) 2. AkhmadZaimi (djimie@gmail.com) 3. SofiahNaning (snh@ittelkom.ac.id) 1 3/24/2010
2. Background Routing Management Multi-services multi-treatment single network (best effort & QOS aware) MPLS Network as IP backbone has capacity constraint and routing mechanism constraint (SPF). The new routing management scheme is proposed using Minimum Interference Routing Algorithm (MIRA). Research’s objective increased availability and accessibility Speedy (BE service) to optimize network resource usage 3/24/2010 2
3. Basic Concept of MPLS-TE In MPLS-TE, headend LSR should be accomplished with network topology and other auxiliary information. The required information: bandwidth, TE metric, max. bandwidth, max. bandwidth reserveable, unreserved max. bandwidth and administrative group of links. 3/24/2010 3 MPLS-TE: FIB Construction
4. Basic Concept of MIRA Least cost routing, from 1 9: 1 – 4 – 5 – 9 (cost=4) 2 8: 2 – 4 – 5 – 8 (cost=4, but interfere 1-9 path) 3 8: 3 – 4 – 6 – 7 – 5 – 8 (cost=8) Terms: interference, minimum interference path, critical link Next issues: Observes critical links of QoS traffic with multi-path consideration Analyze best method to optimize netw. resources for both, BE+QoS service, in term of multi-path. 3/24/2010 4
5. Algorithm (MIRA) Input A graph G(N,L) and a set B of residual capacities on all the arcs. An ingress node a and an egress node b between which a flow of D demand units have to routed Output A route between a and b having capacity of D units of bandwidth Algorithm Compute the maximum flow values for all Compute the set of critical links Csd for all Compute the weights Eliminate all links which have residual bandwidth less than D and form a reduced network Using Djikstra’s algorithm compute the shortest path (a-b) in the reduced network using w(l) as the weight on link l Route the demand of D units from a to b along this shortest path and update the residual capacities 3/24/2010 5
6. Algorithm (Enhancement of MIRA) 3/24/2010 Akhmad Zaimi 212080001 6 Input A graph G(N,L) and a set B of residual capacities on all the arcs. An ingress node a and an egress node b between which a flow of D demand units have to routed Output A route between a and b having capacity of D units of bandwidth Algorithm Identify Ingress Egress Node including bw. allocation Compute weight of Ingress Egress Pair: Compute the candidate paths for all Compute the set of critical links Csd for all M1: all links, both MP and BP, are critical link M2: if B(l) < D then l = Csd, resource allocated for both MP and BP M3: if B(l) < D then l = Csd, resource allocated only for MP
7. Algorithm (Enhancement of MIRA) 3/24/2010 Akhmad Zaimi 212080001 7 Compute the weight of critical link Compute network cost (M1, M2, M3) Include link capacity as a factor of network cost (adopt OSPF cost) Eliminate all links with B < D and form a reduced network Use Djikstra’s algorithm to compute the shortest path (a-b) in the reduced network based on network cost Route the demand of D units from a to b and update the residual capacities
10. Routing Mechanism Nodes route traffic relying on least cost routing mechanism. On previous topology, source to destination routing would be as the following table. 3/24/2010 10
11. Analysis and Discussion Modeling Routing Management System Input of MIRA: G(N, L, B); B = residual capacity (Mbps) on each links 3/24/2010 11
13. Modeling Routing Management System Size of Demands for Each Ingress Egress Pairs (Mbps) Weight of Ingress Egress Pairs 3/24/2010 13
14. Modeling Routing Management System The number of candidate paths (main path and backup) is about 251 candidate paths of IP VPN. Then, analyze critical links within all of candidate paths. And here is the result of computation using 3 methods. 3/24/2010 14
15. Highlight of M3 Result Critical link regards to M3 computation 3/24/2010 15 Network cost regard to M3 computation
16. Analysis and Discussion Output Analysis Result path of routing management Then, to prove the interference path produced by SPF algorithm. 3/24/2010 16
17. Output Analysis 3/24/2010 Akhmad Zaimi 212080001 17 Interference Analysis of SPF algorithm: From 13 2 passes through some critical links (hop 3 and 4) From 13 3 passes through one critical links, which is hop 2.
22. Conclusion Implement and enhance MIRA as routing management platform reduces the occurrence of traffic interference and thus network congestion should be minimized. The network optimization (through critical link analysis) would be achieved by managing routing of traffic in the network. Routing management with MIRA is proficient to leverage network performance, even throughput, received packet and delay 3/24/2010 22
PT. TELKOM offers Speedy as internet broadband(BE) solution while provides IP VPN(SLA) for enterprise customers, in the same network.
Best effort depicts of Speedy traffic characteristicQoS aware (sensitive traffic) depicts of IP VPN traffic characteristic
Best effort depicts of Speedy traffic characteristicQoS aware (sensitive traffic) depicts of IP VPN traffic characteristicInterference The interference only take place if the decreasing maximum flow caused by other traffic flow in the same channel. Minimum Interference PathAn explicit route that maximize the minimum maxflow between other ingress egress pair [9].Critical LinkA group of links to the characteristics whenever an LSP is routed over those links the maxflow values of one or more ingress egress pair decreases (denoted by αsd).