Multi-Protocol Label Switching (MPLS)
MPLS – short for Multi-Protocol Label Switching – is a now-aging network routing system that transfers data between nodes using labels that denote predetermined pathways instead of network addresses that refer to the nodes themselves. This article explains how MPLS works, its types, and the core architecture.
What Is MPLS (Multi-Protocol Label Switching)?
MPLS – short for Multi-Protocol Label Switching – is defined as a now-aging type of network routing system that transfers data between nodes using labels that denote predetermined pathways instead of network addresses that refer to the nodes themselves.
Since its inception in the 1960s, the internet has evolved in more ways than was ever imagined. Amazingly, the internet is still changing, bringing us closer and closer to newer technologies yet undiscovered. Data transfer over the internet has as well evolved. Data transfer is perhaps the most critical function of the internet in connecting millions of computers worldwide.
Traditionally, the standard Internet Protocol (IP) and the Transfer Control Protocol (TCP) have regulated how data packets are moved from one point to the other. In this protocol, each router must make an independent decision about every tiny bit of data packet and determine where the network should send it. Multi-Protocol Label Switching was created to circumvent this bottleneck in data transfer across the internet.
Understanding multi-protocol label switching
Multi-Protocol Label Switching or MPLS is a technique used to route and direct traffic in communication technology that uses labels in place of addresses to handle data flow from one router to the other. Ideally, these addresses identify endpoints for each data packet. However, labels do not focus on the destination but instead on routes and pathways that have already been established.
MPLS is a networking technology that directs traffic consisting of data packets along networking routes but through the shortest path described on the labels.
Multi-Protocol Label Switching is one of the Internet Protocol (IP) routing techniques that can work on numerous packets covering more than one network protocol and, as such, is referred to as a multi-Protocol system. Multi-Protocol Label Switching, therefore, supports technologies such as the Asynchronous Transport Mode (ATM), Frame Relay, DSL, etc.
The MPLS transfer protocol primarily controls the forwarding of packets over a private Wide Area Network (WAN), for example, a company with several remote outlets or branches connected to the main center. It resolves the issue of slow data transfer and downtime when using the internet but remains a scalable and protocol-independent technology.
When comparing Multi-Protocol Label Switching with other data transfer methods, MPLS is a technology that increases the speed at which data flows across a network. This is simply because the need for looking up complex routing tables at every node has been eliminated. Previously, each node in the local internet mesh served as a router determining the path for incoming packets by searching through complex tables.
Multi-Protocol Label Switching was initially released in 2001 by the internet engineering task force (IETF). It released both the architecture of the technology and its label stack encoding. MPLS performed similarly to the ATM switch as a faster routing technique than the conventional method. MPLS, however, did not have the setbacks ATM had. MPLS also has the advantage of out-of-band control and maintenance of traffic engineering.
How Does MPLS Work?
Multi-Protocol Label Switching works by addressing incoming packets to their destination based on the information written on their labels. It does not try to guess the address but uses labels to find an established bandwidth for the data packet.
MPLS works in a manner that is slightly similar to IP routing techniques. When a regular router receives an incoming data packet, the only information on the packet is the destination IP address without further details on the routes or manner in which the network should transport the packet. In MPLS, the label contains information about the routes the data packet should take. This eliminates the cumulative delay by routers in ‘thinking’ of the best possible course.
When a data packet enters a Multi-Protocol Label Switching network, it is given a specific forwarding Class of Service (CoS), also called Forwarding Equivalence Class (FEC). The class of service forms a part of the label, showing what type of information is contained in the data packet, be it real-time data like VoIP or emails. With this label, the routers can reserve the fastest paths with the least latency to highly sensitive real-time information like Voice over Internet Protocol (VoIP) and video conferencing.
When a data packet enters an MPLS network, the entry node is called a Label Edge Router or ingress node. The class of service is then added, specifying the type of information in the packet and its priority level. In MPLS, there are predetermined, unidirectional pathways linking routers across the network; the Label Switched Path (LSP). Networks can only forward data packets after the LSP has been established and the ingress node has encapsulated the packet in the LSP.
Other nodes within the network are called the label switch routers, which are transit nodes ensuring continuous data flow. The information in the packet label guides the transit nodes, and stops are minimized. After passing through the ingress nodes and transmit nodes, the last router is called an egress node, and it removes the label so the packet address can be read and delivered to the destination.
The MPLS uses a networking protocol that is somewhat a combination of Layer 2 (data link layer) and Layer 3 (IP layer) of the Open Systems Interconnection (OSI) model. This is why MPLS is generally considered a layer 2.5 networking protocol, having features from both for data transfer across a network. Its functionality is enabled by the following
Components of the MPLS label:
- Label/label value: It is a 20-bit long field containing the information routers read in directing the data packet.
- Traffic class field: This is a 3-bit long part of the label used to set the Quality of Service and explicit congestion notification.
- Bottom of the stack: Labels can be stacked on top of each other, and the topmost label is in charge of delivery and is replaced by other labels underneath it until the transfer is complete. The last label in an MPLS header is referred to as the bottom of the stack.
- Time to Live (TTL): It is an 8-bit long label that decreases in value each time the packet hops and therefore limits the packet’s lifespan.
Types of MPLS
MPLS technology can be of three types. These are:
Types of MPLS
1. Layer 2 point-to-point
Layer 2 point-to-point is a type of MPLS suitable for companies that need high bandwidth connections connecting a few locations together while maintaining cost-effectiveness. Examples of practical use of layer 2 point-to-point include several network operations with their primary network infrastructure built using Ethernet and layer 2.
Layer 2 point-to-point is an excellent alternative to high bandwidth leased lines. It is not bound by internet protocol and can send data running in the Local Area Network (LAN) directly to the WAN without needing routers to change the packets to be compatible with layer 3 of the OSI model. Here are its pros and cons:
- Pros: With this type of MPLS, the need to manage complex routing tables has been eliminated. Also, it is cost-effective, as WAN connections can be directly linked with layer 2 switches, eliminating the need for expensive routers.
- Cons: It is challenging to get circuits of less than 10Mbps in bandwidth as providers only sell high bandwidth circuits. Further, it does not support point-to-multipoint connections.
2. Layer 2 Virtual Private LAN Services (VPLS)
Layer 2 Virtual Private LAN Services (also known as Layer 2 VPLS) is now becoming more sought after for its ability to provide Ethernet services. Layer 2 VPLS combines the Multi-Protocol Label Switching with the Ethernet and extends the benefits to end customers and carriers.
For over 20 years, LAN has predominantly used Ethernet switching for connectivity, while the carrier network relies on internet protocol routing. Internet protocol not only provides internet access but also provides virtual private network (VPN) access.
Ethernet, however, has continued to be widely used over various bandwidths because it requires little technical knowledge and remains more affordable. Ethernet is now the infrastructure of choice in both LAN and WAN. Virtual Private LAN Services (VPLS) is an ideal protocol that can provide its users with Multi-Protocol Label Switching and Ethernet, therefore diverting all the traffic in Layer 2 directly to the wide area network. In addition, VPLS remains simple, easy, affordable, and highly scalable. Here are its pros and cons:
- Pros: It provides a transparent interface that does not require investment in hardware such as routers to upgrade bandwidth. Traffic is labeled with a MAC address as opposed to an IP address, and like all switched networks, Layer 2 VPLS offers lower latency periods than a router network will offer. Configuration and deployment are straightforward, even for newly added sites.
- Cons: Layer 2 VPLS is still being used only in some parts of the world and has not attained global reach. Therefore this limits the applicability of any feature. The absence of routers as part of the hardware infrastructure places the layer 2 VPLS at higher risk of storm damage. Monitoring is complex due to a lack of visibility from the providers.
3. Layer 3 IP/VPN
Layer 3 IP/VPN is a type of MPLS network most suitable for large enterprises covering multiple branches over a vast land mass. This includes corporations with offices spread across the globe, industries located in more than one country, etc.
Layer 3 IP/VPN is a service that is naturally a continuation of the ATM and legacy frame relay models. Layer 3 IP/VPN transports data packets based on labels attached as the packets enter the ingress nodes. Therefore, it is highly suitable for companies that are merging for easy scalability and rapid deployment.
It is also a good fit for companies migrating from the ATM to IP or from the inflexible frame relay to IP, and also for those preparing for voice and data convergence. Layer 3 IP/VPN makes it possible for all the sites in the network to have a blanket class of service prioritization based on the type of traffic (e.g., VoIP). Here are its pros and cons:
- Pros: Layer 3 IP/VPN is highly scalable and helpful when considering fast deployment. It supportsquality of service (QOS) for differentiation of traffic types. Unlike an ATM, it does not need permanent virtual circuits yet provides the same services.
- Cons: Changing the network settings like QOS takes time and involves sending requests. Layer 3 IP/VPN is not suitable for small businesses. It offers only IP services, and must convert data from layer 2 to layer 3 before you can use it on the network.
Architecture of MPLS
MPLS architecture comprises a combination of 2 OSI layers – i.e., the second and third layers. This means that in an MPLS network, there are unique steps that a data packet must follow to get it across the MPLS domain. These steps include:
- Label creation and distribution must be done based on the FEC and dispersed among the routers with LDP protocol.
- Creation of tables at each router using the Label Forwarding Information Base (LFIB). The LFIB can be regarded as analogous to the routing table employed in the IP network.
- Label switched path creation.
- Label insertion/table lookup of data packets entering the ingress router.
Packet forwarding occurs at every router by swapping the labels until the bottom stack label is reached at the egress router. The primary architectural point of Multi-Protocol Label Switching is that one can add labels carrying additional information to data packets for transfer above what the routers previously had to use.
Apart from this, you must understand the five elements of MPLS to grasp the architecture of the network.
1. Ingress Label Edge Router (LER)
The ingress label edge router is located on the periphery and indicates a point of entry for the data packet from its source. Ingress label router imposes a label and forwards the packets to a destination. Therefore, the ingress edge router is responsible for initiating the packet forwarding operation and does this just after setting up the label switched path (LSP) and assigning proper labels.
2. Forward Equivalence Class (FEC)
The Forward Equivalence Class is a group of data packets related to one application that is forwarded in its switch path, applying the same treatment and across the same route. Therefore, all the packets of that class bear the same service requirement. Each type of data traffic is given a new forward equivalence class, which is done immediately when the packet enters the MPLS cloud.
3. Label Switch Router (LSR)
The Label Switch Router is a part of the MPLS that exchanges inbound packets with outbound ones. It also performs functions such as label removal or disposition, label addition or imposition, and label swapping. In label swapping, the label switch router replaces the topmost label in a stack with the value of an outgoing label. This router also separates data streams from the access network into the core of the MPLS, into different FECs.
4. Label Switch Path (LSP)
The Label Switch Path (LSP) is a direct pathway in the Multi-Protocol Label Switching (MPLS) enabled network that is used by a packet moving from its source to the destination. LSP is a unidirectional path that allows packets to move in only one direction. The packet passes through several intermediate routers between the origin and destination.
A labeled switched path is necessary for every MPLS network for data transfer to occur. A typical scenario involves a data packet coming in from the ingress node (LER) and migrating through different nodes through the shortest possible path, using an established LSP before getting to the egress node.
5. Egress Label Edge Router (LER)
Like the ingress LER, the Egress Label Edge Router (LER) is a router located on the MPLS network’s periphery. It serves as a point of exit for data packets that have arrived at their destination. Therefore, it removes labels (label disposition) and forwards the IP packet to the final destination. The egress LER uses a bottom-of-stack indicator to guide its function. This means it will only dispose of a label if the label on top of the stack is identified as a bottom label.
Multi-Protocol Label Switching is also separated into the control and forwarding planes:
- MPLS control plane: The responsibility of the control play is to create the label switched path. The LSP is then used for sharing the routing information through the routers and also integrates the data, creating the LFIB.
- MPLS forwarding plane: The forwarding plane directs packets throughrouters based on their labels. It uses the information in the LFIB.
Takeaway
While MPLS remains foundational to network infrastructure, its usage is waning. According to a 2021 study by Telegeography, implementation of MPLS decreased by 24% between 2019 and 2020. During this time, the adoption of SD-WAN increased, speaking to the growing preference for more agile and flexible software-based technologies.
On the other hand, MPLS involves expensive but highly reliable infrastructure which promises excellent performance, especially for real-time data transfers. As a result, certain enterprises may want to hold onto their MPLS investments and have them co-exist with new technologies.
