Onto the next Core Architecture. MPLS.
1.1 Describe service provider architectures
1.1.a Core architectures (Metro Ethernet, MPLS, unified MPLS, SR)
Cisco service provider service delivery is basically based on MPLS. For that reason this is probably the more important concept to know about, but it will be hit on for basically every other point of the exam so I think they mean to understand it at a high level and be able to compare it to other technologies.
MPLS stands for Multi-Protocol Label Switching. It is an overlay technology which means it allows for traffic to be transported between two points on an underlying network that may be based on a different technology. MPLS is designed to support transporting multiple protocols, such as Ethernet, IP, TDM, or SONET.
Why might a Service Provider want to use MPLS? Using an overlay technology like MPLS overcomes most of the downsides of a Metro Ethernet platform as discussed in the last post. An ISP can use MPLS to create separate overlay paths for each customer service back to the core so there is none of the shared resource constraints of Ethernet. Also, MPLS running over a Cisco routed network allows us to use the IGP of our choice and deploy in Mesh or Ring typologies as needed. OSPF or ISIS are very traditional for this purpose. There are also several add on technologies such as Label Distribution Protocol (LDP) and Traffic Engineering (TE) that let the provider more simply operate the network and provide resiliency.
These are all pretty high level concepts for MPLS, but I think that’s the point of this exam topic. Cisco expects us to know why MPLS is better than Metro Ethernet in most cases. I will say there is one downside that still exists to this day which is cost. Generally speaking, MPLS capable hardware costs more than the comparable Metro Ethernet solution. In Cisco’s world a service provider MPLS network relies on ASR9000 and ASR900 series routers for the most part.
As a practical note, service providers probably wind up using a combination of MPLS and Metro Ethernet for end to end service delivery. MPLS runs to some point in the network where traffic is then handed to a Metro Ethernet platform for last mile delivery. Some providers do run MPLS all the way to the customer edge, but many stop at some higher aggregation point and use MPLS to segment Metro Ethernet networks to reasonable sizes.
The only key point to understand at this level is that MPLS packets are forwarded based on a label rather than an IP address or MAC Address. There are special ways in which routers forward MPLS packets which is similar to IP routing and Ethernet forwarding combined. The final important detail to know is that during the forwarding operation the router never even looks at the IP or Ethernet headers. In fact, the payload might not even have an IP or Ethernet header. The whole point is to forward only based on the label.
- Provider Edge Router (PE) – This is the router that sits closest to the customer equipment and is where the MPLS network stops.
- Provider Core Router (P) – Basically any other router within the provider network
- Customer Edge Router (CE) – Also just known as CPE, this is the customer side of the connection
- Autonomous System Boundary Router (ASBR) – In traditional MPLS this is another place where the MPLS network stops. In unified MPLS you can transit these
- Route Reflector (RR) – BGP route reflector, to be studied in detail later
I am far from an expert on implementing MPLS (though I plan to be one by the time I take the exam). The points I have made here are a high level overview of what MPLS is and why service providers use it. I’ll follow up this post with the rest of 1.1.a to discuss unified MPLS and Segment Routing (SR).