MPLS and Carrier Networks

Upon completion of this course, you will be able to:

• Explain the components and basic structure of a carrier packet network including core, provider edge, access and customer edge,
• List three ways carrier packet services are better than dedicated lines or ISDN for wide-area networking,
• Define a Service Level Agreement, Class of Service and traffic profile,
• Define a traffic class and explain what a virtual circuit is, and what virtual circuits are used for,
• Differentiate between a reliable and unreliable Class of Service and what must be done to accommodate the latter, and briefly explain connection-oriented and connectionless communication modes,
• Identify the steps involved in communicating voice in packets, and what transmission characteristics are critical to call quality,
• Explain the terminology specific to MPLS including Label Edge Router, Label-Switched Path, Forwarding Equivalence Class, Next-Hop Label Forwarding Entry, including defining the meaning and purpose of a label, and identifying where the label is placed in the headers,
• Trace the flow of a message transported by TCP in IP packets over an MPLS network,
• Identify the benefit of MPLS from the user's point of view,
• Explain what Differentiated Services are, and how MPLS labels can be used to implement Diff-Serv, and an alternative,
• Explain how and why MPLS can be used to achieve service integration,
• Show how MPLS can be used to aggregate traffic,
• Explain what exactly someone means when they say "MPLS service" and explain why "IP service with a service level agreement" would be a more accurate term, and
• Identify two differences between MPLS service and Internet service, and the pros and cons of each.

Lesson Notes

Technical Background and Introduction

A router is a device than relays packets from one circuit to another on a first-come, first-served, packet-by-packet basis. Knowing which circuit to relay the packet to is the routing part of the story, also called packet switching and packet forwarding.

Routers implement bandwidth on demand by not reserving a fraction of the capacity of the connecting circuit for each device (channelizing), but instead giving each device the possibility of using the full capacity of the connecting circuit - when there is something to transmit.

Since devices generate traffic in bursts, and normally have nothing to transmit, many more devices can be connected to the circuit using bandwidth on demand instead of channelizing.

This is called overbooking or oversubscribing. It is done on purpose, to give the users the highest performance for the lowest cost.

In this course, we will take the same idea and apply it again at the carrier network level: replacing the dedicated lines between customer locations from the simple framework of the previous course with bandwidth on demand service from a carrier between the customer locations.

This brings the same benefit to the customer as it did to individual devices in the previous course: the highest performance for the lowest cost.

All of the carrier's customers in a city are given access to the same high-speed intercity circuits, with the possibility of transmitting to other cities at full line speed - but only when they have something to transmit.

This is called a packet network service provided by a carrier.

This type of service is used by businesses (including government, organizations and other carriers) to implement cost-effective, flexible, high-speed packet communications between specific locations.

It is, of course, the same story for the packet networks built by carriers and connected to form the Internet.

Overbooking connections between routers implements bandwidth on demand for the users. In a commercial environment, for banks, government, other carriers, the next questions would be "how MUCH bandwidth on demand" and "how often do we get that"?

This course focuses on the technologies for managing traffic and guaranteeing performance for commercial users on an overbooked network. After covering the basic architecture, we'll discuss the "guarantee", called a Service Level Agreement, and then the critical ideas of virtual circuits and traffic classes, implemented in the past with ATM and Frame Relay, and most importantly, how MPLS today uses these ideas to implement different service levels, aggregation, integration and convergence.

We'll complete the discussion by understanding what sales people mean when they say "MPLS Service" and compare that to Internet service.

List of Lessons

Lesson 1. Course Introduction
The first lesson recaps the concepts of packet switching and bandwidth on demand, then provides an overview of the topics that will be covered, the course objectives and a description of each lesson. This lesson is available for free above and provides both a walkthrough of the course and a sample of the quality of the course graphics, text and presentation.

Lesson 2. Carrier Packet Network Basics
The fundamental concepts of packet switching and bandwidth-on-demand or overbooking, the physical components involved in using a carrier packet network service including Customer Edge, types of access circuits, the Provider Edge and the network core - and why PE equipment is sometimes deployed at the customer premise. This lesson is completed with a roundup of the benefits of packet services over dedicated lines and circuit-switched connections.

Lesson 3. Service Level Agreements: Traffic Profile and Class of Service
How performance is specified, measured, guaranteed and controlled on an overbooked bandwidth-on-demand network - the Service Level Agreement where the network guarantees specified transmission characteristics, sometimes called a Class of Service, on condition that the customer stays within a defined traffic profile … and what happens to out-of-profile traffic.

Lesson 4. Virtual Circuits
The fundamentals of virtual circuits, an essential part of all packet communication networks. We'll cover the concepts of traffic classes, virtual circuits, virtual circuit IDs and the fundamental principles of operation that are common to all technologies, including MPLS, and how virtual circuits are a powerful traffic management tool.

Lesson 5. QoS Requirement for Voice Over IP
Packet network services were originally designed for datacom. In this lesson, we’ll understand how voice is packetized, carried over a packet network, then reconstructed at the far end - and the transmission characteristics necessary for voice quality.

Lesson 6. MPLS
IP has emerged as the standard for packets that will carry all traffic. Since IP provides a connectionless network service, additional protocols are required to implement virtual circuits to enable management and prioritization of traffic. The choice for IP is Multi-Protocol Label Switching (MPLS). The concepts are the same as other virtual circuit technologies X.25, Frame Relay and ATM… but the jargon is changed. We'll begin by identifying MPLS components, jargon and basic principles of operation.

Lesson 7. TCP/IP over MPLS
In this lesson, we'll revisit tracing the path of a file download from server to client, this time over an MPLS network. This will reveal a significant advantage of MPLS-based network services compared to Frame Relay in the user-network interface. We'll also discuss the "M" in MPLS, noting how MPLS can be used to carry frames for VPLS in addition to the usual IP packets.

Lesson 8. Differentiated Classes of Service using MPLS
Here, we'll examine how classifying traffic and mapping classes onto virtual circuits can be a Quality of Service (QoS) mechanism to implement multiple Classes of Service on a packet network. This is sometimes referred to as differentiated services or Diff-Serv, i.e. providing a different Class of Service for each application: VoIP, IPTV, email, web surfing and others.

Lesson 9. Integration and Convergence using MPLS
In this lesson, we'll see how virtual circuits and traffic classification can be used to combine all of the types of communications of a business or organization onto a single access circuit. This idea is sometimes called convergence, though service integration is a more accurate term. It results in a large cost savings compared to one access circuit for each type of communications.

Lesson 10. Managing Aggregates of Traffic with Label Stacking
Here, we'll understand how MPLS labels can be stacked. In other words, virtual circuits carried over other virtual circuits, and how this is implemented to aggregate traffic for both routing and prioritization reasons - both on access circuits and in the network core.

Lesson 11. MPLS Services vs. Internet Service
This lesson completes the course with a discussion of terminology used in sales and marketing of MPLS services, and how that translates to reality. We will use a quiz question-and-answer format to understand the difference between Internet service - which might be called an SD-WAN, and what sales brochures call "MPLS service" … and what exactly an "MPLS service" is.

Different Traffic Classes