This article was contributed by Rajesh Jhamb who is the founder of OutreachMantra.
Have you ever ordered something online and then tracked the package to see it make strange and illogical stopovers all over the nation?
This is similar to how IP routing works on the Internet. An internet router only receives a packet that contains the destination IP address. The packet is not given any instructions on how to get it to its destination, or how the packet should be handled along the way.
Every router must make a separate forwarding decision based on the network-layer header of each packet. Every time a packet reaches a router, it must “think” about where to send it next. This is done by the router referring to routing tables.
This process is repeated for each hop until the packet reaches its final destination. All those hops, and all those individual routing choices result in poor performance when it comes to time-sensitive applications such as videoconferencing or Voice over IP (VoIP).
What is MPLS?
MPLS is a proven networking technology used by enterprise networks since over 20 years. MPLS is different from other protocols which route traffic according to the source and destination addresses. Instead, MPLS routes traffic using predetermined labels.
MPLS is used by businesses to connect remote offices to the data centers or headquarters of their companies.
How MPLS works
When a packet first enters a network using MPLS, it is assigned to a specific Forwarding Class of Service (CoS), also known as a Forwarding Equivalence Class (FEC). This CoS is indicated by adding a short bit-sequence (the label) at the end of the packet. These classes often indicate the type of traffic that they carry. As an example, a company might name the classes real-time (voice and video), critical (CRM and vertical apps), and best efforts (Internet and email). Each application is placed into one of these categories.
This would ensure that the quality of real-time applications like voice and video is maintained. Other routing protocols make it impossible to separate traffic by performance.
The labels are the key point in the architecture. They allow you to add additional information beyond what routers were able to do before.
What is the MPLS layer 2 and 3?
MPLS has been a source of confusion as to whether it is a Layer 2 service or if it's categorized under Layer 3. MPLS does not fit into the OSI seven layer hierarchy and is often classified as Layer 2.5. MPLS has the advantage of separating forwarding mechanisms and the data-link service. MPLS is able to create forwarding table for any protocol.
MPLS routers create a label-switched route (LSP), which is a predetermined path for routing traffic in an MPLS Network, based on criteria defined in the FEC. MPLS forwarding is possible only after the LSP has been created. LSPs have a unidirectional nature, meaning that the return traffic will be sent via a different LSP.
An ingress MPLS Router, located at the network edge, adds an MPLS Label when an end-user sends traffic to the MPLS network. MPLS Label is divided into four parts:
- The Label
Labels contain all the information needed by MPLS routers in order to determine the destination of the packet.
Quality of Service (QoS), which is used to determine the priority of the packet, uses experimental bits.
Bottom-of-Stack is used to inform MPLS routers that they have reached the end of their journey. There are no labels left. This is usually a router that's an egress.
The number of hops a packet can produce before it's thrown away.
Click here to learn more about MPLS questions.
MPLS Pros and Cons
MPLS offers scalability and performance. It also allows for better bandwidth utilization.
MPLS does not offer encryption. However, it is a virtual network that is separated from the Internet. MPLS can be considered a safe transport mode. It is also not susceptible to denial of service attacks that could impact IP-based networks.
A MPLS connection costs much more than a standard Internet connection. MPLS was also designed for companies that have many remote branch offices located geographically across the nation or world, where most of the traffic is routed back to the enterprise data center. MPLS is no longer optimal because businesses are now rerouting a large portion of their traffic to cloud providers.
MPLS networks in the cloud
As businesses migrate to the cloud, hub-and-spoke models based on MPLS become inefficient. They route traffic through corporate headquarters which are central choke points. Direct traffic to the cloud is more efficient. The increased use of mobile apps, cloud services and video has also led to an increase in bandwidth requirements. MPLS services can be difficult to scale up on demand.
MPLS was an innovative technology for its time. However, there are better technologies to address the network architectures of today. SD-WAN was designed with cloud connectivity as a priority, which is the reason so many businesses are replacing or augmenting MPLS networks.
MPLS vs. SD WAN
SD-WAN refers to the application of Software Defined Networking concepts (SDN) to the WAN. SD-WAN edge devices are deployed to apply rules and policies in order to route traffic according the best possible path.
SD-WAN can be used to route any kind of traffic including MPLS. SD-WAN has the advantage that a central enterprise WAN traffic architect can easily apply policies to all WAN devices.
With MPLS on the other hand, routes must be carefully predetermined and, once they are in place, it is not as simple as clicking a button.
Once an MPLS network has been deployed, real-time performance is guaranteed. SD-WAN routes traffic in the most efficient way, but there is no guarantee of performance once IP packets reach the Internet.
SD-WAN costs less to deploy and run than MPLS. Lightyear's WAN Connectivity Pricing Guide estimates that the average monthly recurring costs of a 100 Mbps MPLS connections are $1,277. SD-WAN, however, is only $300.
Is MPLS dead?
Most network professionals view MPLS and SD WAN as either/or options. SD-WANs are gaining momentum at the expense of MPLS. MPLS usage fell by 24% between 2019 and 2020. In the same period, enterprises began using SD-WANs in greater numbers, from 18% up to 43%.
SD-WAN is destined to kill MPLS, right? Zeuz Kerravala of Network World believes that the two technologies are compatible and MPLS can play a different role. Many small and midsize businesses have shifted to an all broadband WAN model and can abandon MPLS.
Enterprises that have invested in MPLS networks will probably adopt a hybrid strategy. They will use MPLS to run legacy applications on the network and offload Internet traffic (like cloud) to SD-WAN. Hybrid WAN networks are not a new concept for businesses, as they already use hybrid computing, storage and applications.
MPLS will play a key role in connecting point-to point locations such as large regional offices, retail stores with point-of sale systems, regional manufacturing plants, and multiple data centres. MPLS is a good fit for real-time applications such as telepresence. SD WAN, as Verizon, a MPLS provider, points out, can help you make the most of your MPLS connection. SD-WAN promises to dynamically route network traffic to meet quality of service requirements. It can use your MPLS connection for this.
Enterprise WAN architects must make a decision based on a risk/reward analysis between the more reliable, but more expensive performance of MPLS and the less reliable, but cheaper performance of the Internet. Improvements in other networking technologies and protocols have made internet traffic more reliable, but for some there will always be a place for the ultra-high-reliability of MPLS. After all, no one wants to be caught in the crosshairs if the CEO's videoconference with the branch office employees breaks off mid-sentence.