Optical/IP Networks

Multiprotocol Label Switching (MPLS)

Before reading this you may find the following tutorials useful:
Protocol Basics, Formatting for Transmission, Internet Protocol (IP) The primary function of multiprotocol label switching (MPLS) is to perform what is known as “flow aggregation.”

In an IP (Internet protocol) network, constant streams of traffic between two points in the network must be divided into many IP packets that are required to carry the information. Each packet will have its own header, containing the source and destination IP address information. Every IP router in between the source and destination must interrogate every IP packet header in order to route each individual packet according to its destination address. When there is a large amount of traffic all destined for the same destination from the same source, it is quite a waste of resources to have to study every single IP packet header. MPLS aims to recognize such packets and add a special label identifying them as a unified flow of information. Intermediate routers would then be able to quickly see the labels without having to delve into the IP header, and forward them rapidly towards their destination. Through this improved performance of the network, it is then possible to offer quality of service (QOS) guarantees for certain traffic – a marked improvement over IP’s best-effort service.

The journey begins from a source at the edge of the network, where the Layer 3 protocol, probably IP, generates packets. Located here is a label edge router (LER) that handles the entrance and exit of all information in the MPLS network.

MPLS is independent of the Layer 2 and Layer 3 protocols, so it could be used, for example, in an IP-over-ATM or IP-over-Ethernet system. Something known as the forward equivalence class (FEC) identifies flows and classifies them in terms of any QOS requirements.

The LER then assigns a label to the flow of traffic, which it embeds in the Layer 2 protocol (e.g., ATM or Ethernet) header of each frame. This is a compact, fixed-length label which is readily accessed by intermediate nodes from the Layer 2 header, rather than regular IP routers, which must first strip off Layer 2 information to get into the IP header. This rapidly speeds up the routing process.

A protocol known as the label distribution protocol (LDP) ensures that the other MPLS nodes throughout the network know which labels correspond to which routes. Routes can be discovered initially by the same protocols that establish routes for IP routers – such as BGP (border gateway protocol) or OSPF (open shortest path first). The nodes in between the source and destination LERs – known as label switch routers (LSRs) – can now switch the labelled packets of information rapidly through the network on what are known as label switched paths (LSPs).

An MPLS Network The flow of information now safely reaches its destination, another LER, which can now strip off the labels and various headers to get down to the data enclosed. And so the data has travelled through the network far more quickly than previously possible, and with a guaranteed QOS that was not previously possible with standard Internet protocol.

As a couple of final notes, you may sometimes hear MPLS referred to as “tag switching.” This is the name of the protocol initially created by Cisco, and this tutorial applies equally well with the word “label” just replaced by “tag.” MPLS is designed to be an internationally standardised protocol and not vendor-specific. You may also hear the phrase MPlambdaS where the word “lambda” is referring to the switching of wavelengths through the network – another more futuristic application of the protocol.

Key Points

  • Performs “flow aggregation” to speed up the transfer of lots of data travelling between two specific points
  • Removes the need for IP header interrogation of every packet at every intermediate node
  • Label edge routers (LERs) control traffic entering and exiting the MPLS network
  • Forward equivalence class (FEC) identifies and classifies traffic flows according to any QOS requirements
  • LER assigns a label to each packet of the flow in the Layer 2 header for quick access by intermediate label switch routers (LSRs)
  • LSRs quickly switch the flow, based upon the label, through a label switched path (LSP) – they have been informed of the route matching each label through the label distribution protocol (LDP)
  • At the destination is another LER which strips off the labels to leave the data

mgozitn 12/5/2012 | 3:06:10 PM
re: Multiprotocol Label Switching (MPLS) Hi

You can see that the writer is very knowledgeable but when you prepare yourself to here more good news the article is already finished; as he started mentioning LDP (CR-LDP I guess) I was at least expecting him to hear more about RSVP-TE, MPLS-TE , IS-IS, and MGPLS at least. Bandwidth scalability for larger networks: seeing that weGÇÖve now got CRS-1 and T1600 from Cisco and Juniper Respectively. I hope the writer will once again comeback and finished what he had already started.

Themba Mgozi
halibo 12/5/2012 | 1:35:12 AM
re: Multiprotocol Label Switching (MPLS) Tony, Himanshu,

I aggree. When you read this you think it's all about switching packets faster. We all know there is more.

Anyways you can see that LRs focus is on optical from the list of beginners guides :)

himanshu_ashwani 12/4/2012 | 10:19:16 PM
re: Multiprotocol Label Switching (MPLS) This article provides very basic information. Although this may be justified but including hyperlinks in the article which point to other useful and detailed information can be very useful.
tony1athome 12/4/2012 | 8:14:42 PM
re: Multiprotocol Label Switching (MPLS) This discussion would be more complete if it mentioned traffic engineering and RSVP. As it stands, it talks only about things that beginners won't be interested in.


Sign In