Traffic Engineering and Routing by jbw10297

VIEWS: 36 PAGES: 26

									Traffic Engineering and Routing

           Hansen Bow
                  Topics
•   Traffic Engineering with MPLS
•   Issues Concerning Voice over IP
•   Features of Netscope
•   QoS Routing for High-Speed Networks
•   QoS Routing for Multimedia
Traffic Engineering with MPLS:
            Methods
• MPLS – multiprotocol label switching
• Constraint-based routing
• Enhanced link state IGP
    MPLS-multiprotocol label
          switching
• forwarding scheme
• at ingress of MPLS network, IP packets are
  classified, tagged, and routed
• at next router, tag is used to determine
  destination
• before leaving, tag is removed
     Constraint-Based Routing
• computes bounded routes
• reservable bandwidth of a link is an
  approximation
• can be done online or offline
     Enhanced Link State IGP
    (interior gateway protocol)
• distribute link information
• flood network to obtain information
     Deploying MPLS System
          (GlobeCenter)
• Statistics Collection
• Deploy LSP (label-switched path) with
  bandwidth constraint
• Periodic update of LSP Bandwidth
• Offline Constraint-Based Routing
        Voice over IP Issues
• modify capacity management and routing
  methods in IP to support IP telephony
  – delay less than 300ms
  – loss rate <1%
• First Model: RSVP
• Second Model: voice service uses Virtual
  Private Network
             RSVP Routing
• Shortest Path First
• Shortest Available Path First
• Widest Available Path First
       Virtual Private Networks
•   interconnects telephony switches
•   Direct Path Only
•   Success to the Top
•   State-Dependent Routing
•   Approximate State-Dependent Routing
Direct and Alternate Routing
          Policies
Differently Routed Calls with
     Trunk Reservation
Routing Policies with Integrated
        Services Model
• not much difference between SPF, SAPF,
  WAPF
• fewer blocked calls for a given network
  capacity because of better sharing of
  network capacity
        Features of Netscope
• traffic measurement and network modeling
• provide global views of configuration and
  useage data
           Utility of Netscope
•   realizing customer SLAs
•   tuning parameters of network components
•   unite configuration
•   experiment with possible solutions to
    variable complex traffic
                      Data
• Network components
• Modeling Traffic
• Routing
  – multiple shortest paths
               Visualization
•   Objects
•   Statistics
•   Traffic and Links
•   Changing Routes
   QoS Routing Issues for High
        Speed Networks
• Goals
  – satisfy QoS requirements for admitted
    connection
  – global efficiency
• Classes
  – Source routing
  – Distributed routing
  – Hierarchical routing
                   Routing
• Collection of State Information
  – local and global state
  – aggregated global state
                          Routing
• Finding Feasible Path
   – Unicast
      •   link optimization
      •   link constrained
      •   path optimization
      •   path constrained
   – Dijkstra’s algorithm
      • http://www.cs.uwa.edu.au/undergraduate/courses/230.300/read
        ings/graphapplet/graph.html
   – Multicast
      • Steiner Tree
             Routing Strategies
• Source routing
   – centralized problem, loop-free
   – need global state, computation overhead
• Distributed routing
   – routing response faster, scalable
   – need global state, more messages, loops
• Hierarchical routing
   – scales well, computation shared
   – imprecision because aggregate state, complicated with
     constraints,
            Future Directions
•   Efficient Routing Algorithms
•   Routing with imprecise state information
•   Multipath Routing
•   Rerouting
   QoS Routing for Multimedia
• Metric Selection
  – Efficient algorithms must exist for path
    computation
  – Reflect basic information of network
  – orthogonal
• Multiple Metrics
  – additive, multiplicative, concave
Bandwidth and Delay as Metrics
• finding path subject to two or more additive
  and multiplicative metrics is NP-complete
• only feasible combination is bandwidth and,
  for example, delay
  Path Computation Algorithms
• Source Routing
• Hop-by-hop
  – compute best path to every destination
  – shortest-widest path is free of loops
                       References
• “NetScope: Traffic Engineering for IP Networks,” A. Feldmann, A.
  Greenberg, C. Lund, N. Reingold, and J. Rexford, IEEE Network,
  Mar./Apr. 2000, pp. 11-19
• “Traffic Engineering with MPLS in the Internet,” X. Xiao, A. Hannan,
  B. Bailey, and L.M. Ni, Ibid., pp. 28-33
• “Capacity Management and Routing Policies for Voice over IP
  Traffic,” P.P. Mishra, and H. Saran, Ibid, pp. 20-27
• “Quality-of-Service Routing for Supporting Multimedia Applications,”
  Z. Wang, and J. Crowcroft, IEEE Journal on Selected Areas in
  Communications, Vol. 14, No. 7, Sept. 1996
• “An Overview of Quality of Service Routhing for Next-Generation
  High-Speed Networks: Problems and Solutions,” S. Chen and K.
  Nahrstedt, IEEE Network, Nov./Dec. 1998, pp. 64-79

								
To top