mpls and traffic engineering

MPLS and Traffic Engineering

CMPT771 - 2008 Spring Liang Zhou Jiang (Jeff)

Outline

Traffic engineering in IP Networks

Multiprotocol Label Switching

MPLS Applications

Traffic Engineering with MPLS


Traffic Engineering in IP Networks

Coping with Internet growth Network architecture Capacity expansion

Increase the number of circuits Increase the bandwidth of existing circuits Increase the capacity of the core routers Add more core routers

Traffic engineering Wait until QoS becomes ubiquitous and leverage it for admission co

ntrol Better utilize all available bandwidth in the core


BGP Topology


Internet Traffic Engineering Motivation

Architecture paradigms and simple capacity expansion are necessary, but not sufficient, to deliver high quality Internet service under all circumstances.

Definition The aspect of Internet network engineering that addresses the issue of performance

optimization of operational networks. Application of technology and scientific principles to the

measurement Modeling Characterization Control of Internet traffic

Application of knowledge and techniques to achieve specific performance objectives. Goal and purpose

Reliable and expeditious movement of traffic through the network Efficient utilization of network resources Planning of network capacity


Conventional IP Technologies

Advantages Highly distributed and scalable

Disadvantages Doesn’t consider the characteristics of offered traffic and network

capacity constrains when making routing decisions Inadequacy of measurement functions The limitations of intra-domain routing control functions

Consequences Poor network resources allocation Some subnet resources become congested Some subnet resources along alternate paths remain underutilized


MPLS Idea: leverage layer 3 interior routing protocols (OSPF and IS-IS) to calculate

shortest paths to all possible destinations, but then assign a sequence of labels/tags along each path.

Label Switched Path (LSP) and LSP tunnel


MPLS – Cont’CMPT771 - 2008 Spring Liang Zhou Jiang (Jeff)

MPLS – Cont’ Traffic parameters

Adaptivity attributes Priority attributes Preemption attributes Resilience attributes Resource class affinity attributes Policing attributes

Components of the MPLS Traffic Engineering Model Path management

Path selection Path placement Path maintenance

Traffic assignment Partitioning function – partition ingress traffic according to some principle of division Apportionment function – allots the partitioned traffic to established LSP tunnels according to some principle of

allocation Network state information dissemination

Extending conventional IGP to propagate additional information: Maximum link bandwidth, maximum allocation multilplier default traffic engineering metric, reserved bandwidth per priority class, and resource class attributes

Network management


MPLS Applications Short cut routing

BGP Next Hop A MPLS Short Cut to BGP Next Hop

Tunnel Restoration Planned Head End Reroute Capability Fast Reroute

Integrating MPLS and QoS Integrated Services Differentiated Services IntServ Meets DiffServ and MPLS at the Internet Core


MPLS – Cont’


MPLS – Cont’ L-LSP’s for Mapping DiffServ to MPLS E-LSP’s for Mapping DiffServ to MPLS Major components of Traffic Engineering with MPLS

User interface for articulating traffic engineering policy in terms of constraints to conventional SPF

IGP component which is composed of traffic engineering extensions to IS-IS and OSPF

Signaling component which is based on traffic engineering extensions to RSVP or CR-LDP

Traffic Engineering Policy


Traffic Engineering with MPLS Administratively Specified Explicit Path Selection Resource Class Affinity Label Switched Path Adaptivity Label Switched Path Bandwidth Reservation Priority Label Switched Path Preemption Load Distribution Across Parallel Label Switched Paths Label Switched Path Resilience


Conclusion

The resiliency and adaptability of the Internet is unparalleled in the history of communications. The Internet, with its growing suite of open and standardized protocols, is the clear winner in the inevitable convergence of private line, voice, video, and outsourced data services. MPLS is only the latest entrant in this remarkable evolution. MPLS when combined with traffic engineering deliver a formable tool for meeting the rigid requirements of differentiated services by leveraging the strengths of IP routing, the proven scalability of terabit routers, and the mechanisms for end to end QoS.


References "A Framework for MPLS", Ross Callon, George Swallow, N. Feldman, A. Viswanathan, P. Doolan, A. Fredette, 09/22/1999. (180569 bytes) "Multiprotocol Label Switching Architecture", Ross Callon, A. Viswanathan, E. Rosen, 08/27/1999. (145481 bytes) "MPLS Label Stack Encoding", Dino Farinacci, Tony Li, A. Conta, Y Rekhter, Dan Tappan, E. Rosen, G. Fedorkow, 09/13/1999. (46971 bytes) "Extensions to RSVP for LSP Tunnels", Der-Hwa Gan, Tony Li, George Swallow, Lou Berger, Vijay Srinivasan, Daniel Awduche, 09/29/1999. (105164 bytes) "MPLS Support of Differentiated Services", Bruce Davie, Pasi Vaananen, Liwen Wu, Francois Le Faucheur, Pierrick Cheval, Ram Krishnan, Shahram Davari, 10/11/1999. "Applicability Statement for Extensions to RSVP for LSP-Tunnels", Alan Hannan, Daniel Awduche, X Xiao, 10/05/1999. (17395 bytes) “MPLS Optimized Multipath (MPLS--OMP)”, Curtis Villamizar, February 25, 1999 "OSPF Optimized Multipath (OSPF-OMP)", Curtis Villamizar, 02/25/1999. (90622 bytes) “IS-IS Extensions for Traffic Engineering”, Henk Smit, Tony Li, May 1999 “OSPF Extensions for Traffic Engineering” Derek M. Yeung, February 1999 “RSVP Label Allocation for Backup Tunnels”, Robert Goguen, George Swallow, October 1999


mpls and traffic engineering

Documents

partitioned traffic

traffic engineeringcmpt771

reserved bandwidth

available bandwidth

path lsp

maximum link bandwidth

congestedsome subnet

ip networkscoping