The History Of Multiprotocol Label Switching

Published Date: 02 Nov 2017

Introduction:

MPLS stands for Multiprotocol Label Switching Protocol. This protocol contains data packets that are attached with labels. MPLS enables the router to forward packets through the network by referencing the packet label content instead of using the destination IP, therefore the packets are forwarded using label switching. When packets are forwarded using MPLS an additional layer of label forwarding is provided to the IP routing table which makes the routing of packets quicker and separates the traffic. The router will not check the packet content but will only read the label attached to forward the packet. MPLS is found at a layer called "Layer 2.5" which is located between layer 2 (Data link layer) and layer 3 (Network layer).

MPLS enables networks to carry all different types of traffic such as IP, VOIP traffics, and it also unifies many networks into one unified network which results less costs. MPLS was designed to provide a service that can carry any type of data through the MPLS network. Before the implementation of MPLS, there were few other protocols that were used to switch labels such as: Frame relay and ATM. But then there are differences between these protocols and why the MPLS is becoming more popular:

- Cost reduction when combining networks: MPLS is able to combine Frame relay, ATM, IP, and Ethernet networks into one network which reduces costs.

- Voice, video, and data services: MPLS enables these services which in one service called "Triple Play" on a common Backbone.

- Flexibility and high availability of MPLS additional to its multiple services support makes it the most common protocol used today since it offers many features that other protocols don’t, as well as the implementation of MPLS reduces many costs.

The main reason of the success of the MPLS protocol nowadays is that it can carry many protocols using the label on the packet such as: IPV4, IPV6, and many layer 2 protocols. MPLS contain a feature called Any Transport over MPLS (AToM) which enables the transport of Layer 2 frames within an MPLS network.

Architecture:

The MPLS architecture enables the protocol to perform Label switching, and as mentioned earlier MPLS is located between layer 2 and layer 3, MPLS architecture contains few on the functions of these layers to forward packets: Switching in Layer 2 and routing in layer 3. The difference between MPLS and other network technologies is how the labels are allocated and the capability to transport these labels attached to the packet.

MPLS architecture contains 2 planes that enable it to forward and control the labels:

Data plane: to forward a packet, the MPLS refers to a database located in a label switch.

Control plane: responsible to bind information to forward the label through label switches.

C:\Users\Shouba\Desktop\01fig03.gif

Figure 1: MPLS performing IP routing.

MPLS Label:

http://www.hep.ucl.ac.uk/~ytl/qos/images/mpls_headerlabel_01.gif

Figure 2: MPLS label header.

The figure above display the label MPLS label header that is added to each packet, this label stack contains four fields that contain information about the label. The first 20 bit field contains information about the value of the label and what operations are assigned to the label. The second field (3 bits) is used for experimental purposes and for quality of service.

However the next 1 bit field is usually called the Bottom of stack which contains a value of zero if it is not the bottom of the stack, if it is the bottom of the stack then it contains the value 1, the value is set to one when a router requires more than one label attached to the packet in order to transmit it through the MPLS network, then these labels are packet into one stack and the lowest label in the stack is the bottom of stack and contains the value 1 in the third field of the label header.

Finally the last field which is an 8 bit field is called the Time to Live field, this field operates similarly to the TTL field found in the packet which determines for how long the label could be transmitted in the network without reaching the destination, and by that the packet is retransmitted again towards the destination, the purpose of the TTL field is to pervert network congestion so that the packet won’t travel into the network looking for the destination for too long.

Label switch router:

A label switch router (LSR) is any router that could forward (transmit) and receive labeled packets in an MPLS network, in other words a label switch is any router that supports MPLS. However, in an MPLS network there are three types of Label switch routers which are Ingress (Edge LSR), Egress (Edge LSR), and Intermediate LSR’s. However all three LSR’s can do: POP, Push, and Swap operations. Edge LSR’s are the routers that are allocated at the edge of an MPLS network unlike the intermediate LSR’s which are usually located between the edge LSR’s.

Ingress label switch router: this type of LSR adds a label into a packet that they received and after adding the label, it transmits it back to the network, this router only add the label to unlabeled packets that it receives.

Egress label switch router: After receiving a packet that is attached with a label, the egress LSR separates the label from the packet and transmits the packet through the network link.

Intermediate label switch router: This type of LSR receives a labeled packet, however this label swaps the label and transmit the packet back according to the label.

In general, I would like to explain the link between all LSR’s in an example: A PC sends a packet containing a destination IP through a basic MPLS network, the ingress router receives the unlabeled packet from the sender PC and attaches a label to it containing value of 6 and sends it the intermediate router which will swap the label implemented by the ingress router and adds a new value such as 3, then the intermediate router will transmit the packet using the new value towards the Egress label which will remove the label and use the destination address located in the packet header to transmit it to the destination PC.

One of the differences between these LSR’s that only Edge networks conduct a router lookup to forward the packet, on the other hand, intermediate LSR conduct basic label lookup and switching to switch packets from one edge LSR to another edge LSR.

Figure 3: LSR’s in an MPLS network.

In figure 3, a basic MPLS network is shown containing both edge LSR’s (Ingress and Egress) and intermediate LSR’s which are the ones located between edge LSR’s. However, LSR’s use a path called LSP (Label Switching Path) which is an MPLS network path used to transmit labeled packets through the MPLS network. Usually the ingress LSR is the first point of that LSP and that makes the Egress LSR the last point of that LSP. Referring to the figure above, it shows the direction of the path is from left to right which determines the LSP for that label, however labels can also be transmitted from the other side of the network (Right to left) but then that will be on a different LSP.

Label distribution MPLS:

As mentioned in the example earlier that the first label(s) is added by the ingress LSR and from then LSP for that label is determined, however being transmitted through the network the only changes implied to the label stack is the top label where it gets swapped at each hop. Intermediate LSR is responsible for swapping the top label of the received packet with another label and send the packet. The egress LSR receives the labeled packet and removes the label stack from the packet and forwards the unlabeled packet through a link towards the destination address.

Switching labels with new labels requires the agreement of adjacent routers on the prefix used, in order for the routers to communicate and agree on the prefix used for the label switching, they need to use a label distribution protocol. However, there are two methods if distributing labels:

Piggyback the Labels on an existing IP Routing Protocol:

This method enables the label to be carried by an IP routing protocol and that works by extending the protocol. By using this method, the synchronization between the routing and label distribution is always initiated which means that no label will be added to the packet if the prefix is not found. This terminates the need of other protocol by the LSR to perform the label distribution.

Running a separate Protocol For Label Distribution:

In this method, label distribution in LSR is handled by a separate protocol which leaves the routing protocol to be independent and dispense prefixes. There are some disadvantages to this method, the main one being the increase of protocols used by the LSR to distribute labels. One of the separate protocols used to distribute labels is the Label Distribution Protocol (LDP).

LER (Label edge routers):

LSR (Label switch routers)

LSP label switched paths

CPE (Customer premises equipment)

LDP (Label distribution protocol)

http://www.youtube.com/watch?v=H7KQcNRgGEk

VPN

Trouble shooting and maintainance