How MPLS (Multiprotocol Label Switching) Works
Enterprise networks are often tasked with bringing a bunch of smaller networks together. This could include connecting branch offices, different buildings on a campus, or remote monitoring stations. There are many ways it comes together.
When that is the goal, you have a few options for how you actually design the wide area network, and one powerful solution is multiprotocol label switching (MPLS).
What Is MPLS?
Let’s start at the beginning. What is MPLS? It’s a routing technique that focuses on hardware controls to make networking more efficient. When used well, it can improve data rates and stability, especially for traffic that is deemed more important.
Typically, MPLS is used at the enterprise level because it is designed for linking multiple networks together into a wide area network (WAN).
The technique first debuted in 1994 when Toshiba released the original Cell Switch Router. In the decades since, MPLS has remained a core method for routing in large networks.
How Does It Work?
Faster, more efficient networks sound nice, but how does MPLS actually achieve these benefits? That’s best understood by comparing MPLS to traditional IP networking.
When you aren’t using a special method, networks route information through individual data packets. Each packet has an IP address on it, so the router knows where the packet is coming from and where it needs to go. With that information, the router can choose a pathway to get the packet to its destination, and that choice will be based on available resources. Each packet will typically be sent along the path that has the most available bandwidth at the moment of transmission.
MPLS completely changes this method of routing by making one simple change. It allows each packet to hold a header, and that header consists of tags that provide a little more information about the data in the packet. Tags can identify packets as carrying video data or sound data or many other types of data.
With this extra information, MPLS routers can create a traffic hierarchy. They can prioritize packets that are marked as more important in the hierarchy, and those packets will be given the fastest or slowest routes, depending on the rules set up by the IT administration.
As a result, the packets that are deemed more important will have lower latency (because they always get the preferred pathways), and less important traffic will take the longer, slower routes. This allows a network to group traffic according to latency needs.
As an example, video streaming (which needs low latency to operate smoothly) can get those preferred lanes. Meanwhile, downloads for software updates can run on the slower lanes because it’s not a real-time application. The update can’t run until all of the packets are there anyway, so small delays don’t matter as much.
When used well, this hierarchical approach can improve user experiences across the network.
Fitting MPLS into the OSI Model
MPLS offers some networking advantages that become clearer when we delve into the OSI model. As a quick reminder, this is the theoretical networking model that breaks digital communication into seven specific layers.
Typically, routing happens in Layer 3 while core networking happens in Layer 2. MPLS is considered a Layer 2.5 protocol. While it doesn't actually create a brand new networking layer in order to operate, MPLS designs hybridize the two layers in order to create this label-driven routing mechanism.
To keep a long story short, this hybridization enables MPLS designs to consolidate a lot of Layer 2 resources. This helps with networking efficiency overall, extending the benefits gained from investments into MPLS.
SD-WAN vs MPLS
Clearly, MPLS is a great way to think about enterprise networking, but it’s also more than 25 years old. These days, there are modernized solutions that offer competitive advantages over MPLS, and it’s important to understand those options.
Primarily, you can replace MPLS with SD-WAN. This is a software-defined (hence the “SD”) approach to WAN networking, and it’s basically a more powerful, adaptable way to achieve the same benefits as MPLS. With SD-WAN, you can use software controllers to tag packets and create data hierarchies. But because it’s all at the software level, it can rapidly change the rules and adjust routing in real-time.
Also, the software approach is a lot cheaper. With MPLS, you have to create hardware lanes that are then utilized by the packet labels. That costs money. SD-WAN creates virtual lanes that use existing networking infrastructure more efficiently without the need for tons of equipment. As a result, an SD-WAN can speed up your large networks for much less cost.
All of that said, MPLS still shows up in a lot of networks, and there’s one clear reason for that. Despite the extra costs, the rigidness of MPLS is an advantage in some scenarios. Since it’s difficult to change the hierarchy in an MPLS network, any system that benefits from a static hierarchy will prefer MPLS to SD-WAN.
An example will clarify. Say the WAN monitors natural gas lines to look for leaks. The most important component of that network is the system that can actually flag a gas leak. That system always needs to have the highest priority, so it is better served by MPLS than an SD-WAN that might lower that system’s priority from time to time.
To generalize, if a real-time application is mission-critical within the network, it’s worth considering MPLS to ensure that application is always at the top of the hierarchy.
Additional Learning Center Resources