MPLS Explained: What Is Multi-Protocol Label Switching? (Simple Guide)

Hello! This is Lazarus at telecom.Tech, where  telecom and networking technologies... are simply   explained! Today we'll be simply explaining  multi-protocol label switching or MPLS.  Now this is the first video in a new series of  videos about MPLS. If you want to be informed  

about every new video that comes out so that you  can follow along, please subscribe and enable   notifications. And if you like this video, please  like this video. What is MPLS? To be honest with   you, I had a really hard time when I was learning  about MPLS. It's a technology that's quite  

different from the more traditional IP routing  and Layer 2 switching that we learn about, and   for this reason it was difficult for me to get my  head around. There are so many different ways to   implement it depending on what you want to achieve  and it interacts with many other features and  

protocols in order to fulfill its function. So as  I was thinking about how to approach MPLS in this   series, the thing that really stuck in my mind was  the fundamental principle that I always try to use   with all of my videos: I'll try to explain MPLS  in a way that I wish I had been taught when I was  

learning about it. So hopefully this approach will  work out well for you too! So what is MPLS? In its   most fundamental form, it's a high-performance  network protocol used to efficiently direct   data between nodes. It's a protocol that you  would most often find within the core network  

of an ISP or other large geographically dispersed  networks. In contrast to more traditional routing   based on network addresses, MPLS uses labels to  identify established paths between endpoints on   the network. MPLS is able to encapsulate  packets of all types of network protocols,  

and that is where the "multi-protocol" part of  its name comes in. Indeed MPLS can encapsulate   a whole range of technologies including IPv4,  IPv6, Ethernet, ATM, frame relay, and even lower   layer technologies such as T1/E1 lines as well  as DSL. So in this sense, MPLS can be considered  

a tunneling protocol, although the term "tunnel"  is not as commonly associated with it. Instead,   MPLS is viewed more like an underlay network  capable of transporting various payload types   and supporting multiple overlay technologies.  Now because of its nature, MPLS is sometimes  

referred to as a "Layer 2.5" protocol because it  operates in a manner that seems to be somewhere   between the traditional Data Link Layer which  is Layer 2 on the OSI model, and the Network   Layer which is Layer 3 on the OSI model.  And as you you will see later in this video,  

when the MPLS header is actually appended to a  packet, it's actually placed between the Layer 2   and Layer 3 headers of the data. Now having said  all of this, what is MPLS really used for? It's   not a technology that you would find deployed  in a small or medium-sized enterprise network,  

and unless an organization is extensively  large and owns its own WAN infrastructure,   you would not typically see it even on large  enterprise networks. MPLS is a technology   that is most often deployed within geographically  large ISP networks. Its purpose is to interconnect  

multiple remote sites. An ISP will typically have  many customers and an MPLS network can be used   to interconnect the remote locations of those  customers. It provides an ideal infrastructure   to be able to securely and efficiently separate  the traffic from multiple customers, but also  

allow that traffic to leverage the same single  MPLS infrastructure. One other point I would   like to make here is that MPLS's primary purpose  is not to connect a premises to the Internet. In   other words, MPLS is not a WAN access technology  or a last mile technology like cable or DSL or  

fiber to the premises. MPLS may be deployed over  these last mile technologies, but MPLS itself does   not constitute a WAN access technology. So the  connection here for example to Customer 2 may   be a fiber connection that delivers connectivity  to the MPLS infrastructure, it may be a DSL line,  

it may be cable, it may be a wireless microwave  link... it doesn't really matter. MPLS is not   involved with that last mile but it is served by  those last mile technologies. Now why use MPLS?   Why not just use traditional IP routing to enable  communication between these customer sites? Well,  

there are several advantages that MPLS provides  over IP routing that make it a much better choice,   and arguably the most important benefit is  that on high-capacity high-throughput networks,   label switching is much more efficient than IP  routing. Now to see this, let's take a look at  

how IP routing would be used to route data through  a network such as this. Now imagine that each of   these routers is running a routing protocol such  as OSPF, or EIGRP, or even BGP, and imagine that   there are thousands of customers connected to  this ISP network. That means that each of these  

routers will have created a routing table that  has been populated by potentially thousands of   routing entries. Once a packet reaches this router  here, the destination IP address of the packet   is examined and compared sequentially with every  entry in the routing table until a match is found.  

It may have to go through dozens, hundreds, or  even thousands of entries before it finds the one   that matches. Once that's done, the exit interface  is determined and the packet is sent out of that   interface to reach the next node in the path. At  this point the process is repeated in this node  

with a routing table lookup, a match with the  destination IP address, and the packet is then   sent along its way. Once again, the process is  repeated at each subsequent node until the packet   reaches the intended destination. Now I'm sure you  can see that this process is quite CPU intensive  

at each individual node, and if you have thousands  of customers with thousands of destination   addresses with innumerable packets running  through this high-capacity backbone network,   you can imagine the CPU and memory resources  needed to achieve this at each node. Now label  

switching which MPLS uses is much more efficient  and is thus preferable. Let's see it in action.   When an IP packet arrives at the first router of  the MPLS infrastructure, it is associated with a   specific label based on the specific customer from  which the packet came. The packet is prepended  

with an MPLS header which contains that label.  Now the MPLS header looks something like this,   and as you can see, the header is placed  between the IP header here and the frame header,   right between Layer 2 and Layer 3. As the packet  traverses the MPLS network, each intermediate node  

examines only the label attached to the packet,  not the full IP header, and it uses that label to   make forwarding decisions. That label is looked  up in the forwarding table and the next node   is determined. Every time the packet reaches  another node, a label lookup takes place and  

the next node in the path is determined, and the  packet is forwarded at each node. The label may   also be swapped with a new label for the next  segment of its journey. This process is known   as "label swapping." So all of this continues for  every router that's encountered. This assignment  

and swapping of labels is at the very core of  what MPLS does. Eventually, the packet reaches   the final router at the edge of the MPLS  network. Here the MPLS label is removed and   the packet is forwarded based on its original IP  header. Now the key here is that labels simplify  

the forwarding process by reducing the need for  complex IP header lookups at each hop by relying   on short fixed-length labels. MPLS can improve the  speed and efficiency of packet forwarding in the   network and this is a vital importance for ISP  networks that carry a lot of data and that want  

to ensure that all of their customers can enjoy  high-performance network services. Now once an   MPLS path is established, packets with the same  labels typically follow the same path through the   network and this is unlike traditional IP routing  where packets might actually take different paths  

based on current route availability or changes  in the routing decisions. So all of these   characteristics make MPLS efficient and preferable  for such network infrastructures. So that's my   introduction to MPLS...simple right? We'll go into  more detail in future videos but for now, I hope  

you found this video useful, and if you have,  please click the thumbs-up button! And if you'd   like me to address other related topics please  feel free to let me know in the comments below,   and please subscribe to get updates to newly  published videos. I'm Lazarus at telecom.Tech,  

I hope this video has been helpful for you,  and I'd like to thank you for watching!