VIDEO
HIGHLIGHT
Log InSign up
1.3 The network core
SummaryTranscriptChat

1.3 The network core

Introduction to Network Core

In this section, we will explore the network core and its operations. We will discuss packet switching, forwarding, routing, and the structure of the internet.

Network Core Overview

  • The network core consists of interconnected routers and communication links.
  • The core operation is based on packet switching.
  • End hosts divide application-level messages into packets and send them into the internet.

Forwarding and Routing

  • Forwarding is the local action of moving a packet from an input link to an appropriate output link.
  • It is controlled by a forwarding table in each router.
  • Routing is the global action of determining source-to-destination paths for packets.
  • Routing algorithms compute these paths and per-router forwarding tables.

Analogy: Trip in a Car

  • Understanding the difference between forwarding and routing can be compared to taking a trip in a car.
  • Routing determines the path from source to destination, while forwarding handles local actions at intersections.

Transmission of Packet Bits at a Router

This section focuses on how packet bits are transmitted within a router in the network core. It explains store-and-forward operation and queuing.

Store-and-Forward Operation

  • Packets are transmitted from one router to another by gathering up bits until the full packet has been received.
  • This process is known as store-and-forward operation.

Queuing

  • If packets arrive at a router faster than it can transmit them, queues form in that router.
  • Queuing occurs when work arrives faster than it can be served by the service facility.

Conclusion

The transcript provides an overview of the network core, including concepts such as packet switching, forwarding, routing, store-and-forward operation, and queuing. Understanding these fundamental aspects helps comprehend the functioning of the internet and its infrastructure.

Queuing in British Culture

This section discusses the British culture of queuing and its significance.

Queuing as an Olympic Event

  • The British are known for their exceptional queuing skills.
  • If queuing were an Olympic event, Great Britain would win gold.
  • The British are willing to wait for extended periods, even up to an hour, and still maintain their patience.
  • Understanding the rules of queuing is essential to fitting in with British culture.

Packet Queues and Queuing Delays

This section explains packet queues and the delays that occur when packets have to wait at routers.

Packet Queues at Routers

  • When the arrival rate of packets on a router's input link exceeds the transmission rate of its output link, packet queues form.
  • Packets have to wait in routers instead of being immediately forwarded to their destination.
  • Limited memory in routers can lead to long queues and exhausted memory.
  • In such cases, arriving packets may be dropped or lost at the router.

Comparison: Packet Switching vs Circuit Switching

This section compares packet switching and circuit switching as different network technologies.

Circuit Switching

  • Circuit switching involves establishing a dedicated path (call) from source to destination before data transmission begins.
  • All necessary resources for the call are allocated upfront, ensuring no queuing delays or loss within the network.
  • Link capacity is reserved exclusively for the call during its duration.
  • However, idle circuits result in inefficient use of bandwidth.

Packet Switching

  • Packet switching does not require upfront call setup or resource reservation.
  • Data is transmitted in small packets independently from source to destination.
  • Each user needs a certain amount of bandwidth but is only busy transmitting 10% of the time.
  • Statistical multiplexing gain allows more users to be supported, even if occasional delays and loss occur.

Supportable Users in a Networking Scenario

This section presents a numerical example to determine the number of users that can be supported in a specific networking scenario.

Numerical Example: End Users and Bandwidth

  • Assume a gigabit per second link with N end users.
  • Each user needs to send data at 1/10th of the overall link bandwidth (megabits per second).
  • Users are only busy transmitting 10% of the time; the remaining 90% is idle.

Circuit Switching vs Packet Switching

  • Under circuit switching, each user needs 100 megabits per second, allowing for a maximum of 10 users.
  • Similarly, under packet switching, each user needs 100 megabits per second, supporting up to 10 users.
  • However, if all 35 users are allowed into the system under packet switching, queues will form less than 0.04% of the time.
  • This statistical multiplexing gain justifies allowing more users into the system despite occasional delays and loss.

Conclusion: Packet Switching as a Winner

This section concludes by highlighting the advantages of packet switching and its widespread adoption.

Advantages of Packet Switching

  • Packet switching is simple and efficient for bursty data transmission.
  • There is no need for call setup or resource reservation; hosts can start sending data immediately.
  • Congestion and packet delay can occur occasionally but are outweighed by statistical multiplexing gain.
  • Even modern telephone networks utilize packet switching for carrying data.

The transcript provided does not include any content beyond this point.

New Section

In this section, the speaker introduces the concept of the internet as a network of networks and discusses how different access networks are connected to each other.

The Structure of the Internet

  • The internet is a network of networks, where users at the edge of the network are attached to various access networks such as home networks, mobile networks, and institutional networks.
  • To connect these millions of access networks together and establish end-to-end paths between users, one approach would be to wire each access ISP to every other access ISP. However, this approach is not scalable due to the large number of connections required.
  • Instead, a global transit ISP can be created where each ISP at the edge connects to a backbone network. This allows one access ISP to reach another through this backbone network.
  • Initially, edge networks were interconnected in this way. However, with multiple global ISPs competing for backbone network services, these backbone networks also need to be interconnected with each other. This interconnection between networks is known as peering.
  • Regional networks may form interconnect access points closer to home and connect both with tier 1 providers and with each other. Content providers like Google or Microsoft may also run their own global networks to bring their services closer to end users.
  • The structure of today's internet consists of well-connected large tier 1 commercial ISPs at the center, regional networks that peer with each other and tier 1 providers, access networks at the edge, and content provider networks like Google and Facebook that connect their data centers directly to the internet bypassing tier 1 and regional ISPs.

New Section

In this section, the speaker further explains the structure of the internet, highlighting the role of tier 1 commercial ISPs and providing an example of a tier 1 ISP's network map.

Tier 1 Commercial ISPs

  • The internet is composed of a relatively small number of well-connected large networks known as tier 1 commercial ISPs. Examples include Level 3, Sprint, AT&T, and NTT. These ISPs have national and international coverage.
  • Moving closer to the edge of the network are regional networks that interconnect with each other and with tier 1 providers. At the very edge are access networks themselves, while content provider networks like Google and Facebook connect their data centers directly to the internet.
  • A tier 1 ISP typically consists of multiple points of presence (POPs), which are collections of routers forming a network node. POPs connect to customer networks, other POPs within the same ISP, and peering networks such as other tier 1 providers.

New Section

In this section, the speaker summarizes important concepts covered in the previous sections related to packet forwarding process, store-and-forward networks, queuing delays, packet loss, packet forwarding vs routing, circuit switch vs packet switch networks.

Recap: Network Core Concepts

  • The previous sections covered various important concepts related to network core:
  • Packet forwarding process
  • Store-and-forward networks
  • Queuing delays
  • Possibility for packet loss
  • Distinction between packet forwarding and routing
  • Comparison between circuit switch and packet switch networks

New Section

In this section, the speaker concludes by summarizing what it means for the internet to be a network of networks and hints at the upcoming topic of network performance.

The Internet as a Network of Networks

  • The internet is indeed a network of networks, with a hierarchical structure consisting of tier 1 commercial ISPs, regional networks, access networks, and content provider networks.
  • The speaker concludes by mentioning that the next topic to be discussed will be network performance.
Playlists: Computer Networking: A Top-Down Approach (All Chapters), Computer Networking: A Top-Down Approach - Jim Kurose
Video description

Video presentation: Computer Networks and the Internet: the network core. Core network functions, packet swtiching, circuit switching, the Internet: a network of networks. Computer networks class. Jim Kurose Textbook reading: Section 1.3, Computer Networking: a Top-Down Approach (8th edition), J.F. Kurose, K.W. Ross, Pearson, 2020. See http://gaia.cs.umass.edu/kurose_ross for more open student resources.