Tuesday, April 3, 2012

Multi-Protocol Label Switching (MPLS)

   This article identifies Multi-Protocol Label Switching (MPLS) technology components, describes their functionality, and illustrates the value they provide in Service Provider environments.

       MPLS was initially targeted for Service Provider customers; however, Enterprises have begun to show interest in deploying this technology. This document can apply to large Enterprise customer whose networks resemble Service Provider networks in the following areas:
  • Size of the network
  • Offer "internal services" to different departments within the Enterprise
   MPLS compliments IP technology. It is designed to leverage the intelligence associated with IP Routing, and the Switching paradigm associated with Asynchronous Transfer Mode (ATM). MPLS consists of a Control Plane and a Forwarding Plane. The Control Plane builds what is called a "Forwarding Table," while the Forwarding Plane forwards packets to the appropriate interface (based on the Forwarding Table).
   The efficient design of MPLS uses Labels to encapsulate IP packets. A Forwarding Table lists Label Values, which are each associated with determining the outgoing interface for every network prefix. Cisco IOS Software supports two signaling mechanisms to distribute labels: Label Distribution Protocol (LDP) and Resource Reservation Protocol/Traffic Engineering (RSVP / TE).

MPLS comprises the following major components:
  1.  MPLS Virtual Private Networks (VPNs)—provides MPLS-enabled IP networks for Layer 3 and Layer 2 connectivity. Includes two major components:    1.  Layer 3 VPNs—based on Border Gateway Patrol    2.  Layer 2 VPNs—Any Transport over MPLS (AToM)
  2. MPLS Traffic Engineering (TE)— provides an increased utilization of network bandwidth inventory and for protection services
  3. MPLS Quality of Service (QoS)— buildings upon existing IP QoS mechanisms, and provides preferential treatment to certain types of traffic, based on a QoS attribute (i.e., MPLS EXP).
MPLS VPNs (Layer 3 VPNs)
   Layer 3 VPNs or BGP VPNs have been the most widely deployed MPLS technology. They use Virtual Routing instances to create a separate routing table for each subscriber, and use BGP to establish peering relations and signal the VPN-associated labels with each of the corresponding Provider Edge (PE) routers. This results in a highly scalable implementation, because core (P) routers have no information about the VPNs.

   BGP VPNs are useful when subscribers want Layer 3 connectivity, and would prefer to offload their routing overhead to a Service Provider. This ensures that a variety of Layer 2 interfaces can be used on either side of a VPN. For example, Site A can use an Ethernet interface, while Site B uses an ATM interface; however, Sites A and B are part of a single VPN.

It is relatively simple to implement multiple topologies with router filtering, including a Hub & Spoke or Full Mesh:
  • Hub and Spoke—central site is configured to "learn" all the routes from the remote sites, while the remote sites are restricted to "learn" routes only from the central site.
  • Full Mesh topologies would result in all the sites having the ability to "learn" or import routes from every other site.
    Layer 3 VPNs have been deployed in networks that have as many as—seven hundred PE routers. Service Providers are currently providing up to five hundred VPNs, with each VPN containing as many as one thousand sites. A wide variety of routing protocols are available deploy on the subscriber access link (i.e. CE to PE link). These include Static Routes, BGP, RIP and Open Shortest Path First (OSPF). Most VPNs have been deployed with Static Routes, followed by BGP Routing.

   Layer 3 VPNs offer advanced capabilities, including Inter-AS and Carrier Supporting Carrier (CSC). These provide hierarchical VPNs, allowing a Service Provider to provide connectivity across multiple administrative networks. Currently, initial deployments of such functionality are becoming more widespread.
Download MPLS FLASH PRESENTATION here Full Mesh, Point to Point


Post a Comment