Corporates have always been the first to recognize the merit of technology. Be it the internet or a VPN, the drive behind their implementation was for the sake of making businesses bloom. However, with the successful integration of such technology in our daily lives, the possibility for further improvement is not mere fiction. Now, be it the consumer sector or the corporate world, they both crave better tech to make their lives easier. But, not all technology can be tailor-made for the sake of a generic user’s consumption. No matter how genial the solution is. One such improvement in private networks and routing implementations is the introduction of MPLS in recent years. But what is MPLS? How does it work? Today, we will glean deeper into the workings of a multiprotocol label switch and understand what makes the tech enjoy such popularity in the corporate scenario.
Multiprotocol Label Switching (MPLS) is a data forwarding technology that dates back to the 90s. Created to bypass the use of routing tables, it can improve the speed of a network. Thus, it is a tried and true technology; that increases the speed and controls the flow of network traffic. Toshiba, Cisco, Ipsilon, IBM, and IETF aided the modern-day MPLS to find widespread implementation. The data gets directed through a path via labels instead of addresses. It forgoes the complex procedure of looking up routing tables for every forwarding decision.
Traditionally, when data enters a network, it moves along nodes based on addresses. Thus, the data on a router has to take a path based on routing tables, according to its destination, and make decisions for itself. However, MPLS predetermines the road for every data and assigns each a label to facilitate traffic.
We can conclude that: Multiprotocol Label Switching is a mechanism in a high-performance telecommunication network directing data from one network node to other based on SHORT PATH LABELS instead of LONG PATH NETWORK ADDRESSES. Doing so avoids the complex lookups in a routing table and enhances the traffic flow. Thus it relies on the “virtual roads” between the nodes instead of the endpoints. And since it can encapsulate the packets from different protocols, it is multiprotocol.
MPLS can work with network protocols: such as IP, ATM, T1/E1, and Frame Relay. It uses protocols to create a Label Switched Path (LSP). The LSP acts as the transmission path throughout the MPLS network, aiding the packets along their journey. And while MPLS is a perfect implementation of a private network, we can also find it working at the public ISP. The various application of this tech are as follows:
- TE (Traffic Engineering)
- AToM (Any Transport Over MPLS)
How does MPLS work?
When a data packet enters a network for the first time, it gets assigned a specific forwarding class of service (CoS). We also know Cos as FEC (forwarding equivalence class) has an indicator in the form of a short bit sequence (label) specific to the packet. Now, these classes are often indicative of the type of traffic they carry; For example, real-time, mission-critical, the best effort, etc. Network reserves the paths according to the class. This segregation of traffic based on performance is impossible for other routing protocols, which makes MPLS unique.
Now, the data packets pass through routers onto the predetermined paths known as LSP, which is unidirectional between a pair of routers on an MPLS network. Observe the following actions in each router on the LSP.
- When a packet enters the network through the Label Edge Router AKA Ingress Node, it gets assigned to an FEC depending on the type of data it contains and its destination. FECs can identify packets with similar characteristics.
- Then, based on the FEC, the Ingress Node will apply a label to the packet and encapsulate the data within an LSP.
- Now, the LSP will ensure the safe passage of the data across the network’s transit nodes. Transit nodes are what we call Label Switch routers.
- These routers will continue to direct the traffic according to the label. Although, the in-between stops are not subject to IP lookups but packet labels.
- Finally, at the end of the LSP, the last router will remove the label from the packet, and it will then forward the data according to the IP routing practices. This router is also known as the Egress Node.
Markup of a Label:
It concludes how data travels within an MPLS network and the working of the Multiprotocol Label Switching. However, the markup of a label stack consists of at least four parts, which is detailed below:
- Label Value – It holds the information for routers to determine where the packet should go.
- Traffic Class Field – It sets the Quality of Service (QoS) priority and Explicit Congestion Notification.
- Bottom of stack flag – It is the indicator of the last label in the stack.
- Time-to-live (TTL) Field – It limits the lifespan of the data before it reaches the destination. Or is discarded.
Although, remember that these labels can also be stacked. If they are, then the topmost label will control packet delivery. And with each destination reached, the upper layer will pop, and the one underneath will take over.
Regular IP routing VS routing within MPLS
Regular routing and MPLS routing are a bit different. While one uses network addresses to forward packets, the other relies on LSP labels. Given below is a comparison between routing in an IP network and an MPLS network.
- Switching in the multiprotocol network depends on labels, but an IP network needs an address for forwarding.
- In MPLS, a predefined path exits for packets known as LSP, but IP networks do not specify paths for data.
- A Multiprotocol Label Switching network builds LFIB tables using the LDP protocol. But IP networks rely on routing tables that are time-consuming.
- The MPLS technology sits between layer 2 and 3 of the OSI, but the IP utilizes layer 3 of OSI.
- IP is traditionally slower than the counterpart because MPLS uses labels that avoid overloading the CPU.
- The IP header decides the IP routing process, while MPLS bases its decisions on the MPLS header.
Regular routing or IP routing is the traditional way of forwarding data packets along with a network. Here the process relies entirely on destination IP addresses, and routing lookups get performed at every hop. It is an extensive method that needs every router within a network to obtain thorough routing information to compile routing tables. Thus, IP routing utilizes the hop-by-hop mechanism, relying on the IP addresses as a basis for data packet routing/forwarding.
Any data sent from one computer to another over the internet gets divvied into smaller pieces called packets. Why? Because the communicating language of a computer doesn’t support the transfer of data as a whole. So the data has to be repacked into smaller pieces known as packets and transferred one at a time. Now, these packets have two parts, headers, and data. These headers contain information regarding the source and destination address. And some additional information that helps with routing the packet.
For a packet to reach the destination, routers have to forward it according to the address. The address here is the IP address confirming with the IP network. So a router will first examine the header to determine if the data can travel on the IP network and figure out where to send it. Routers do this by referencing and marinating a routing table. Every router within a network actively and passively contributed to this table. Next, the router forwards the data based on the IP address to the next router inline within the network. The process repeats as data hops from one router to the next reaching the target destination.
It works most of the time. However, with every hop of the data, the routers analyze and refer to the routing table. Thus, the process is slower than MPLS routing.
On the other hand, MPLS routing is relatively faster than any other routing method on the market. Thus, most telecommunication use this network for its efficiency. Although it has been around for quite a while now, it is only in recent years that MPLS saw widespread acceptance. Before that, service providers used IP routing alongside VPNs and layer2 tech to bear consumer satisfaction.
In a typical routing scenario, each router makes decisions independently based on internal routing tables. Even when two packets originate from the same source and head towards the same destination, they may take different routes based on the tables. But, MPLS ensures that: there exists a predefined path for certain types of data to shorten routing time. When a packet reaches the MPLS network, a class gets assigned to it based on the kind of data it carries, the priority level, and the destination. This class or FEC then decides the path for that data packet. The ‘path’ is what we call Label Switching Path. Thus, data packets within the same FEC follow the same LSP, making the process practically faster than traditional routing.
Now, the data packets may have one or more than one label attached to them. However, it includes an MPLS header. This header contains the FEC and LSP information. Thus, when the packet reaches the router, it ignores every other header such as the IP one, and based on the label, it forwards the data. Since the routers only need an MPLS header to function, the network can support most networking protocols.
- LSR receives IP packets and adds an MPLS header.
- Due to this header, the network supports multiple protocols.
- Now, the ‘header’ contains FEC, based on the type of data a packet carries.
- LDP regulated labels decide routing within MPLS.
Significant terms and abbreviations in MPLS
- MPLS – Multiprotocol Label Switching
- LDP – Label Distribution Protocol
- TDP – Tag Distribution Protocol
- RSVP – Resource Reservation Protocol
- LSP – Label Switched Paths
- LSR – Label Switching Router
- MTU – Maximum Transmission Unit
- QoS -Quality of Service
- VPLS – Virtual Private LAN Service
- TTL -Time to Live
- TE -Traffic Engineering
- AToM – Any Transport over MPLS
- FEC – Forwarding Equivalence Class
- LFIB – Label Forwarding Information Base
|Provider Edge (PE) Router
|It is a router at the edge of an MPLS network. It can add or remove labels from IP packets.
|Customer Edge (CE) Router
|CE is a router at the edge of a customer network that sends or receives IP packets from PE.
|Label Switch Router (LSR)
|Routers that can understand labels.
|LSR routers that receive IP packets from CE Routers and add MPLS header.
|LSR routers that swap labels in MPLS header. They can forward labeled IP packets.
|LSR routers that send IP packets to CE routers and remove MPLS header.
|Push, Pop, and Swap
|The action of addition, removal, and swapping of labels; done by the LSR, respectively.
Where does MPLS fall within the OSI 7-layer hierarchy?
Many confuse whether MPLS is a layer2 or layer3 service within the OSI model. But the tech doesn’t seem to quite fit within the hierarchy of the OSI model. Thus, we define it as a layer 2.5 service. One of the benefits of the MPLS network is that it can separate forwarding mechanisms from the underlying data-link services. In simple terms, MPLS can create tables for any underlying protocol that grants it versatility. Any data within the MPLS network travels on a unidirectional LSP. The LSP is predefined, which means that: the returning traffic has to take a different route than the forwarded traffic.
MPLS is a layer2.5 networking protocol. Layer2 carries IP packets over LANs or point-to-point WANs. Meanwhile, layer3 uses internet-wide addressing and routing based on IP protocols. Since an MPLS can combine both and function effortlessly between these two layers, it is widely considered a layer2.5 service with additional features for data transportation across networks.
MPLS label types
Due to its unique structure within the OSI model, MPLS has several label types based on how and where it operates within the OSI at the moment. Such as:
- At layer2: Point-to-point – The label is suitable for organizations that require high data transmission between a few sites.
- At layer3: IP VPN – At layer3, the labels are particularly appropriate for the enormous multi-site undertaking; For example, a corporate store, which sends countless low data transfer destinations or mammoth corporates with worldwide remote workplaces.
- Virtual Private LAN Services: At layer2 – Due to layer3 validation, “Virtual Private LAN” administrations are developing in popularity. They join Ethernet and MPLS networks permitting both users and carriers to benefit.
The architecture of a Multiprotocol Label Switching network
The basic architectural markup of a multiprotocol label switching network is as follows:
1. Forwarding Plane: The forwarding plane takes care of the MPLS and IP data packets by sending them forward accordingly. It relies on:
- FIB: RIB-based FIB routes packets according to the routing data. It process and forwards regular IP packets.
- LFIB: Similarly, LDP creates an LFIB on LSRs (Label Switching Router) to route MPLS packets.
2. Control Plane: It creates and manages the routing and label information.
- RIB: IP routing protocols are responsible for the creation of RIBs within the MPLS network. RIB can select the routes for incoming data packets.
- LDP: LDP is responsible for adding labels. Additionally, it also makes LIB and dissolves LSPs when required.
- LIB: LIB is the perimeter that can control MPLS labels.
Advantages of Multiprotocol Label Switching
Several benefits associated with an MPLS network are as follows:
- Lower congestion
- Improved up-time
- Enhanced bandwidth
- WAN management
- Service level agreements
- WAN protocol support
- Quality of service (QoS)
- WAN routing
- Remote connections
- Common applications
- Enhanced security
Disadvantages of Multiprotocol Label Switching
Similarly, an MPLS network has its drawbacks, despite being an improvement from traditional routing. Such as:
- Lack of total control
- Optimized for point-to-point connectivity only
- Long time to deploy
- Requires streamlining its delivery
- Lack of encryption
- Cloud challenges
Does the advent of SD-WAN signify the end of MPLS?
In conclusion, we would address the concerns related to the emerging technology of SD-WAN and how it is gaining a foothold within the MPLS circle with a headstrong momentum. Honestly, though, MPLS and SD-WANs are two different technologies and not an advancement of others. However, SD-WAN certainly has affected MPLS. While the multiprotocol switching technology isn’t dead yet, the role it plays has changed significantly. Since an SD-WAN can pair up quite nicely with an all-cloud IPT model, many have shifted or are shifting towards a complete SD-WAN architecture.
Although larger corporates and businesses will or do rely on a hybrid approach to get the best of both worlds; But, MPLS will continue as a priority when it comes to point-to-point connectivity. Also, since the MPLS routing can prioritize data traffic, it will further continue to be influential for real-time applications.