Network Layer Chapter 4 Quiz

Chapter 4: Network Layer

• PPT Slide Use: PowerPoint slides are freely available for faculty, students, and readers. Slides can be modified, adapted, and used, but slides used should give credit to the creators of the slides and the copyright owner.

• Chapter Goals: Understand principles behind network layer services, including: networking layer models, services for individual datagrams (forwarding versus routing , how routers work, routing (path selection), broadcast/multicast, instantiation, and implementation in the Internet).

• Chapter 4 Outline:

  • Introduction covering virtual circuit vs. datagram operation
  • In-depth look at IP (Internet Protocol): Datagram format, IPv4 Addressing, ICMP, IPv6
  • Routing algorithms: Link state, distance vector, & hierarchical routing
  • Internet routing: RIP, OSPF, & BGP
  • Broadcast and multicast routing

• Network Layer: Transport segments are transferred from sending host to the receiving host on sending side. Encapsulates segments into datagrams, and on the receiving side, it delivers segments to the transport layer. Network layer protocols are in every host and router. Routers examine header fields in IP datagrams as they pass through.

• Two Key Network Layer Functions:

  • Forwarding: Moving packets from a router's input to the appropriate output, which is the process of getting through a single interchange.
  • Routing: Determining the routes taken by packets from source to destination using algorithms.

• Interplay of Routing and Forwarding: Routing algorithms determine end-to-end path through a network. Forwarding tables determine local forwarding at each router based on the value in the arriving packet's header.

• Connection Setup: A third important function in some network architecture (ATM, frame relay, X.25). Before datagrams flow, end hosts and intervening routers establish virtual connections.

• Network Service Model: The question of a service model regarding "channel" transmission of datagrams from sender to receiver. This section highlights specific services regarding individual and flow datagrams, such as guaranteed delivery and delivery with specific delay requirements. Other options may also include things like in-order datagram delivery, guaranteed bandwidth, and restrictions on spacing.

• Network Layer Service Models: Different network architectures and service models (e.g. Internet, ATM) offer various guarantees regarding bandwidth, loss, the ordering of datagrams, timing, and congestion handling.

• Virtual Circuits: Virtual circuits are analogous to telephone circuits, ensuring performance and actions along a source-to-destination path through call setup and teardown. Each packet carries a VC identifier. Each router along the source-destination path maintains state for the connection. Resources like bandwidth and buffers may be allocated to VCs for predictable service. Virtual circuits are implemented with path from source to destination along with VC numbers for each link path. Entries exist in forwarding tables in routers. The VC number is used instead of destination address on certain links. Signaling protocols are used for setup, maintenance, and teardown of virtual circuits (used in examples including ATM, frame relay, X.25.)

• Datagram Networks: Datagram networks establish no call setup at the network layer, and have no state regarding end-to-end connections. Packets are forwarded using destination host addresses.

• Datagram Forwarding Table: Each router has a forwarding table with destination addresses and the appropriate output link. The table is based on a routing algorithm. Aggregation of ranges of addresses is possible to improve routing efficiency.

• Longest Prefix Matching: In forwarding tables, the longest matching address prefix is used.

• Datagram vs. Virtual Circuit Networks: Datagram networks are more flexible and adaptable, while virtual circuits offer assured services with strict timing requirements.

• Router Architecture Overview: Two key functions include running routing algorithms (like RIP, OSPF, BGP) and forwarding datagrams, where tables are computed and pushed to input ports. This also involves a routing processor, a switching fabric, and input and output ports.

• Input Port Functions: Includes line termination, link layer receiving, data link layer (e.g., Ethernet) handling, and decentralized switching. Using lookup tables, forwarding actions are taken. Speed is a critical component.

• Switching Fabrics: Transfer packets from input to output buffers, with switching rate measured in relation to input and output line rates. Three types of switching fabrics include memory, bus, and crossbar.

• Switching via Memory, Bus, Interconnection Network: Different switching techniques for routers. Memory-based switching is slower than other techniques. Bus-based can have contention issues, and interconnection networks can overcome these limitations.

• Output Ports: Buffering is required at output ports due to the need for scheduling and the potential for datagram loss if the fabric rate is faster than the port's rate. Scheduling prioritization also happens here.

• How Much Buffering? Buffering requirements are often related to round-trip time and link capacity.

• Input Port Queueing: If the rate of input packets exceeds the input queue processing capacity, it may create delays and losses. Head-of-line is a particular concern.

• ICMP (Internet Control Message Protocol): ICMP is used by hosts and routers to communicate network-level information, notably including error reporting, unreachable hosts/networks/ports/protocols, echo requests/replies (used in ping), network-layer “above” IP messages (e.g., congestion control and various other cases.) An ICMP message has a type code, indicating the message type and status, plus 8 bytes related to the packet in question.

• IP Addressing: Introduction, Subnets, CIDR Overview and concepts of IP addressing. Addressing and connections (host, router) to link. Subnets are related to the physical connectivity in routing (devices in same subnet part of IP address can communicate). CIDR uses arbitrary subnet portion lengths. Subnet portions determine whether two or more devices are in the same subnet. A subnet is determined by detaching each interface, creating segmented networks.

• IP Addresses: How to Get One? Techniques for hosts to acquire IP addresses include being hard-coded by a system administrator, using static addresses, and employing DHCP.

• DHCP (Dynamic Host Configuration Protocol): DHCP automates IP address assignment, allowing hosts to dynamically obtain IP addresses from a network server when joining. This provides reusable addresses and eases mobility support. Host broadcasts message request, and server responds with address.

• NAT (Network Address Translation.) NAT allows a local network to use a single IP address globally , without requiring the ISP to assign every device a global IP address. Local networks can reconfigure/change ISP without needing to notify the rest of the network of IP address changes within the local subnet. Local network devices are not addressable or visible by the rest of (the outside) the network, which often provides security benefits. Translation involves replacing source addresses with NAT IP address/port. NAT routers track source-destination address/port pairs in the translation table. NAT routers replace source IP and port and create a unique NAT IP address for each outbound connection. They also do this for inbound connections, changing their original destinations back to the local addresses/ports.

• NAT Traversal Problem: Clients may have trouble connecting to servers if addresses aren't compatible.

• Other changes from IPv4 (in IPv6.) Addresses, check sums are eliminated, and changes are made to ICMP (e.g., new message types like "Packet Too Big"). Multicast support is improved.

• IPv6 (Internet Protocol Version 6) Offers a 128-bit address space to address allocation issues. Specific header changes (for example priority and flow label) affect processing efficiency.

• Transition from IPv4 to IPv6 (Tunneling): A method to facilitate the use of IPv4 based infrastructure when introducing IPv6 as the network standard. This includes encapsulation of IPv6 datagrams within IPv4 datagrams as payloads to transfer among different machines.

• IPv6 Adoption: The transition to IPv6 is slow, with estimates for industry adoption in routers being at ~3%-11% in 2013. The delay of widespread implementation is often attributed to the application level changes required in existing systems (e.g., WWW, Facebook).