(T) Organizations have been demanding virtual private networks (VPNs) instead of costly leased lines from their service providers and, Ethernet and IP services instead of DS1 and DS3 services. MPLS is now the best way for service providers to respond to those customer demands by providing Pseudo Wires services (PWS), Virtual Private LAN services (VPLS) or Layer 2 MPLS VPNs and IP VPNs or Layer 3 MPLS VPNs.
MPLS emerged initially in the late 90s as a traffic engineering technology for the core of the Internet. Since the burst of the Telecom bubble in 2001, service providers have realized that the Internet was not growing so fast anymore and traffic engineering was less needed.
However, the demand from enterprise networks for more Internet Protocol (IP) and Ethernet services and in particular Virtual Private Networks (VPNs), cheaper than leased-lines, has given to MPLS a new life beyond traffic engineering.
MPLS has strongly emerged as the best technology to provide initially IP VPNs also called Layer 3 VPNs and now PWS and VPLS also called Layer 2 VPNs.
For service providers, MPLS VPNs are definitely the revenue generating application. For enterprise networks, VPNs enable them to save costs by moving away from expensive legacy leased lines. In addition, MPLS VPNs enable enterprise networks to switch from legacy Frame Relay (FR) and Asynchronous Transfer Mode (ATM) VPNs to MPLS VPNs in order to accommodate their growth in native IP and Ethernet connectivity.
Service providers generally offer the MPLS service through their provider edge (PE) routers. The enterprise IP networks are connected to their service provider networks through their customer edge (CE) routers.
Fundamentally, MPLS provides a connection-oriented capability to IP, through label switching. MPLS is independent of the layer data link (Ethernet, ATM or FR). In MPLS networks, packets are forwarded based on their forwarding equivalence class (FEC) as they enter the MPLS network.
The FEC to which the packet is assigned is encoded as a short fixed length value known as a label. An FEC is basically a flow of IP packets forwarded over the same path and mapped through the same labels. A label-switched path (LSP) is a simplex layer 2 tunnel like an ATM or FR PVC which defines the path followed by labeled packets assigned.
MPLS can use different distribution label framework. Labels are 32 bits. Present protocols for label distribution include Resource Reservation Protocol (RSVP), Label Distribution Protocol (LDP), and Border Gateway Protocol (BGP). Label distribution is done from downstream to upstream. And, labels can be stacked to provide LSPs hierarchy.
MPLS Label Distribution with RSVP Traffic Engineering (RSVP-TE)
RSVP can be used to create and maintain distributed state information other than pure resource reservations. RSVP is “receiver-oriented”: the receiver of a data flow initiates and maintains the resource reservation used for that flow. Once the path has been marked, the routers are ready for the reservation. RSVP provides downstream-on-demand label distribution and an RSVP flow can be considered as an LSP. RSVP-TE extends RSVP to provide constraint-based routing for LSPs. RSVP has two messages:
- RSVP PATH: the ingress LSR generates an RSVP Path message with a Session and a Label_Request. The following objects can be added but are optional: Explicit_Route (ERO) (a predefined explicit route for the LSP) and Record_Route (RRO) (the ingress LSR can request the list of the LSRs that the LSP tunnel will traverse).
- RSVP RESV: the egress LSR responds with a RESV message containing the Session and the Label.
MPLS Label Switching
MPLS Fast ReRoute
Fast ReRoute provides recovery for node and link failures. The backup nodes, links or series of links or nodes are pre-signaled (protection). Traffic is switched to the backup link around the point of failure (restoration). Two main approaches exist:
- Detour LSPs (also called one-to-one back-up): a “detour” is provided around the node or link failure.
- Facility back-up (also called many-to-one or bypass tunnels): instead of creating multiple “detours”, a single LSP is created for multiple nodes and link failures.
Differentiated Services with Traffic Engineering (MPLS With DiffServ)
When the network load is optimized with MPLS traffic engineering, QoS are more likely to be met. DiffServ can be managed through a traffic engineering framework as defined in the IETF RFC 3270. Different differentiated services can be assigned to different paths. Two approaches exist:
- L-LSP: the DSCP is mapped to the label.
- E-LSP: one FEC for a group of DiffServ services. The DSCP drop precedence is mapped to the EXP field (3 bits) of the MPLS label providing an FEC for up to 8 BAs.
Copyright © 2005-2005 by Serge-Paul Carrasco. All rights reserved.
Contact Us: asvinsider at gmail dot com.