Multi Protocol Label Switching (MPLS)

1. Abstract

Along with the advancement of information and telecommunication technology, the need for a network will increase, especially to connect one network to another, where the two network locations are far apart, so to connect the two to create a faster and better connection, a path called Multi Protocol Label Switching (MPLS) is needed.

As we all know that MPLS is a packet delivery technology on a high-speed backbone network (main network) that combines several advantages of circuit-switched and packet switched communication systems that produce better technology than both. MPLS works on packets with MPLS headers, which contain one or more labels. The MPLS header consists of 32 bits of data, including 20 label bits, 2 experimental bits, and 1 stack identification bit, as well as 8 TTL bits. Labels in MPLS are used for the forwarding process, including the traffic engineering process.

It is hoped that with the MPLS path, a network can be connected and connected easily and it is hoped that the access process can be faster and better.

2. Understanding MPLS 

Multiprotocol Label Switching (MPLS) [1] is a packet delivery technology on high-speed backbone networks that combines several advantages of circuit-switched and packet-switched communication systems, resulting in a technology that is better than both.

Multiprotocol Label Switching (MPLS) [2] is a network architecture defined by the IETF to combine label swapping mechanisms at layer 2 with routing at layer 3 to speed up packet delivery.

Packets in MPLS are forwarded with routing protocols such as OSPF, BGP or EGP. Routing protocols are at layer 3 of the OSI system, while MPLS is between layers 2 and 3. OSPF (Open Shortest Path First) is a link state-based routing protocol (seen from the total distance) after the routers exchange information, a database will be formed on each router. BGP (Border Gateway Protocol) is a router for external networks that is used to avoid routing loops on the internet network.

3. MPLS Header 

MPLS works on packets with an MPLS header, which contains one or more labels. This is called a label stack. The MPLS header can be seen in the image below:

Figure 5. MPLS header

MPLS Header includes:

1. 20-bit label value: A label field that contains the real value of the MPLS label.
2. 3-bit CoS field: A CoS field that can be used to influence the data packet queue and the data packet algorithm as needed.
3. 1-bit bottom of stack flag: If 1 bit is set, then this indicates that the current label is the last label. A field that supports the stack label hierarchy.
4. 8-bit TTL (time to live) field. For 8 bits of working data.

4. Package Encapsulation

Unlike ATM which breaks IP packets, MPLS only encapsulates IP packets, by attaching the MPLS header. The MPLS header consists of 32 bits of data, including 20 label bits, 2 experimental bits, and 1 stack identification bit, and 8 TTL bits. The label is part of the header, has a fixed length, and is the only packet identification mark. The label is used for the forwarding process, including the traffic engineering process. To find out the packet encapsulation in MPLS, see the image Error! Reference source not found. below:

Figure 2. MPLS Packet Encapsulation

Each LSR has a table called a label-switching table. The table contains a mapping of incoming labels, outgoing labels, and links to the next LSR. When an LSR receives a packet, it reads the packet label, replaces it with the outgoing label, and then sends the packet to the next LSR.

In addition to IP packets, MPLS packets can also be re-encapsulated in MPLS packets. So a packet can have multiple headers. And the stack bit in the header indicates whether a header is already at the 'bottom' of the MPLS header stack.

5. MPLS Architecture

MPLS, multi-protocol label switching, is a network architecture defined by the IETF to combine layer 2 label swapping with layer 3 routing to speed up packet delivery. An MPLS network consists of circuits called label-switched paths (LSPs), which connect nodes called label-switched routers (LSRs).

Each LSP is associated with a forwarding equivalence class (FEC), which is a set of packets that receive the same forwarding treatment at an LSR. FECs are identified by labeling.

Figure 3. MPLS architecture

To form an LSP, a signaling protocol is required. This protocol determines forwarding based on the label on the packet. Short, fixed-size labels speed up the forwarding process and increase the flexibility of path selection. The result is a more connection-oriented network datagram.

6. MPLS Network Architecture

1. Classification and labeling of packets. After that, packets will go to the provider (P). From the provider, packets will be forwarded to the core.
2. At the core, packets are forwarded based on labels rather than IP addresses. These labels indicate the class classification (A, B, C, D) and the destination.
3. Removes labels and forwards packets to the receiving end.

Figure 4. MPLS Network Architecture

7. MPLS Cloud

Information :

1. LER : Edge Router Label (label on the side of the router)
2. LSR: Router Switch Label (label on the router switch)
3. FEC: Forward Equivalence Class, forwards packets in the same class.
4. Label : connects a packet in FEC.
5. Stack Labels: various labels containing information about how packets will be forwarded.
6. Label Switch Path: traces packets to direct them to a specific FEC.
7. LDP: Label Distribution Protocol, used to distribute label information between MPLS and network devices.
8. Label Swapping: functions to manipulate labels to forward packets to their destination.

Figure 6. MPLS Cloud

MPLS Network Structure

The MPLS network structure consists of edge Label Switching Routers or edge LSRs surrounding a core Label Switching Routers (LSRs). The basic elements of an MPLS network are:

1. Edge Label Switching Routers (ELSR) 

Edge Label Switching Routers are located on the border of the MPLS network, and function to apply labels to packets entering the MPLS network. An MPLS Edge Router will analyze the IP header and determine the appropriate label to be encapsulated into the packet when an IP packet enters the MPLS network. And when the labeled packet leaves the MPLS network, the other Edge Router will remove the label. Label Switches. This Label Switches device functions to switch packets or cells that have been labeled based on the label. Label Switches also support Layer 3 routing or Layer 2 switching to be added to label switching. The operation in label switches is similar to the switching technique that is usually done in ATM.

2. Label Distribution Protocol (LDP)

Label Distribution Protocol (LDP) is a procedure used to inform the label binding that has been made from one LSR to another LSR in an MPLS network. In the MPLS network architecture, an LSR that is the destination or next hop will send information about the binding of a label to the LSR that previously sent a message to bind the label for its packet route. This technique is commonly called downstream on-demand label distribution.

This new network has several advantages including: 

• MPLS reduces the amount of processing that occurs in IP routers, and improves the performance of sending data packets.
• MPLS can also provide Quality of Service (QoS) in the backbone network, and calculate QoS parameters using the Differentiated services (Diffserv) technique so that each packet service sent will receive different treatment according to its priority scale.

Examples of Using MPLS in Computer Networks

MPLS is commonly used in networks. The following is an example of the use of MPLS in a network that can be seen in Figure 2.6.

Figure 6. MPLS path topology

Information :

For example, if we want to connect the network at Location A with the network at Location C, we can do this in several ways, for example via the routing protocol or via the MPLS line.

1.With Routing Protocol Path

The path from Location A will go to R10 (Router 10) then to R1 (Router 1) then to R2 (Router 2) or to R4 (Router 4) then the path goes to R3 (Router 3) after that to R7 (Router 7) and finally directly to Location C. Routing Protocols that can be used include OSPF, BGP and RIP. The internet path connecting Location A to Location C when using a routing protocol will take longer than the MPLS path because with the routing protocol the path passed is more.

2. With MPLS Line

The path from Location A will go to R10 (Router 10) then to R1 (Router 1) then to R2 (Router 2) or to R4 (Router 4) then the path goes to R3 (Router 3) after that to R7 (Router 7) and finally directly to Location C. Routing Protocols that can be used include OSPF, BGP and RIP. The internet path connecting Location A to Location C when using a routing protocol will take longer than the MPLS path because with the routing protocol the path passed is more.

3. With MPLS VPN

VPN is the same as MPLS path, the difference is only the data sent is encrypted to maintain the privacy of the data. In addition, with MPLS VPN, the path can be shorter by simply connecting the Router at Location A with Location C.

4. Process in MPLS

To find out the switching process that occurs in MPLS, see Figure 2.7.

Figure 7. Switching Process on MPLS Network

1. The working principle of MPLS is to combine switching speed at layer 2 with routing and scalability capabilities at layer 3.
2. The way it works is by inserting a label between the layer 2 and 3 headers on the forwarded packet.
3. Labels are generated by Label-Switching Routers which act as a connector between the MPLS network and the external network.
4. The label contains information about the next destination node where the packet should be sent, then the packet is forwarded to the next node, at this node the packet label will be removed and given a new label containing the next destination.
5. Packets are forwarded in a path called LSP (Label Switching Path).

MPLS Protocol Standardization

There are two protocol standardizations for managing MPLS flows, namely: 

1. CR-LDP (Constraint-based Routing Label Distribution Protocol)

2. RSVP-TE, an extension of the RSVP protocol for design-build traffic.

• An MPLS header does not identify the type of data carried on the MPLS flow.
• If the header carries 2 different types of paths between the same 2 routers, with different treatment from each type of core router, then the MPLS header must specify the path for each type of traffic.

1. MPLS Over ATM 

MPLS over ATM is an alternative to provide IP/MPLS and ATM interfaces in a network. This alternative is better than IP over ATM, because it creates a kind of IP over ATM that is no longer indifferent to each other. This alternative is also better than single MPLS, because it is able to support non-IP traffic if needed by the customer. Figure 2.8. is a picture of MPLS Over ATM.

• Like IP packets, MPLS packets will be encapsulated into AAL 5, then converted into ATM cells.
• The downside of this MPLS over ATM system is that the benefits of MPLS will be reduced, because many of its advantages will overlap with the benefits of ATM. This alternative is not very cost-effective.

Figure 8. MPLS Over ATM

2. MPLS-ATM Hybrid 

Hybrid MPLS-ATM is a network that fully integrates MPLS network on top of ATM core network. MPLS in this case serves to integrate IP and ATM functionality, not separate them. The goal is to provide a network that can handle IP and non-IP traffic equally well, with high efficiency.

The network consists of LSR-ATM. ATM traffic is processed as ATM traffic. IP traffic is processed as ATM-MPLS traffic, which will use VPI and VCI as labels.

The ATM-MPLS cell format is depicted in Figure 2.9.

Figure 9. MPLS-ATM hybrid

The integration of ATM switches and LSRs is expected to be able to combine the speed of ATM switches with the multi-service capabilities of MPLS. The cost of building and maintaining the network is still quite optimal, approaching the cost of an ATM network or MPLS network.

3. Labels and Labeled Packages 

1. MPLS equipment forwards all packets that are labeled in the same way.
2. A label resides in a significant place between a pair of MPLS devices.
3. MPLS labels can be placed in different positions within the data frame, depending on the layer-2 technology used for transport. If the layer-2 technology supports a label, the MPLS label is encapsulated in the original label field.

If the layer 2 technology does not natively support a label, then the MPLS label resides in an encapsulation header.

4. GMPLS

GMPLS (Generalized MPLS) is a vertical convergence concept in transport technology, which is still based on the use of labels like MPLS. After MPLS was developed to improve IP networks, the label concept was used for DWDM-based optical networks, where wavelengths (?) were used as labels. The standard used is called MP?S. However, considering that most optical networks still use SDH, not just DWDM, MP?S was expanded to also include TDM, ADM of SDH, OXC. This broad concept is called GMPLS.

GMPLS is a vertical convergence, because it uses the label switching method in layers 0 to 3 [Allen 2001]. The goal is to provide a network that is overall capable of handling large bandwidth with consistent QoS and full control. And integrated It is expected that GMPLS will replace classic SDH and ATM technologies, which until now are still the most expensive layers in network development. The encapsulation process in GMPLS can be seen in Figure 2.10. below.

Figure 7. GMPLS Encapsulation Process

MPLS Implementation

MPLS is a natural for the IP world. Traffic engineering in MPLS takes full account of the nature of the IP traffic passing through it. Another advantage is that it eliminates the need for technical complexity, such as encapsulation into AALs and the creation of ATM cells, each of which adds delay, adds headers, and increases bandwidth requirements. MPLS does not require these things.

The big problem with MPLS is that it has not yet been supported for non-IP traffic. L2 over MPLS schemes (including Ethernet over MPLS, ATM over MPLS, and FR over MPLS) are in progressive research, but have not yet entered the commercial development stage. What is quite promising is the many alternatives for converting various types of traffic to IP, so that this type of traffic can also be transported over MPLS networks.

Conclusion

MPLS is a packet delivery technology on a high-speed backbone network that combines several advantages of circuit-switched and packet-switched communication systems that produce better technology than both. It can also be said that MPLS is a network architecture defined by the IETF to combine the label swapping mechanism at layer 2 with routing at layer 3 to speed up packet delivery. Unlike ATM which breaks down IP packets, MPLS only encapsulates IP packets, by installing the MPLS header. The MPLS network consists of circuits called label-switched paths (LSPs), which connect points called label-switched routers (LSRs). To form an LSP, a signaling protocol is required. This protocol determines forwarding based on the label on the packet. Short, fixed-size labels speed up the forwarding process and increase the flexibility of path selection. The result is a more connection-oriented network datagram. In MPLS there are two standardizations, namely CR-LDP (Constraint-based Routing Label Distribution Protocol) and RSVP-TE, an extension of the RSVP protocol for traffic engineering.

Figure 0.1

QUESTION

1. How can an MPLS network deliver packets or data to the destination Border Gateway Protocol (BGP)?
2. From the figure 0.1 of the network below, which is the path for MLPS?
3. How does the packet encapsulation process occur in MPLS and try to describe the packet encapsulation in MPLS?
4. What are the advantages of using MPLS in building a network?
5. Why is MPLS Over ATM used in networks?