Computer Networking: A Top-Down Approach, CH4

0.0(0)
studied byStudied by 0 people
learnLearn
examPractice Test
spaced repetitionSpaced Repetition
heart puzzleMatch
flashcardsFlashcards
Card Sorting

1/87

encourage image

There's no tags or description

Looks like no tags are added yet.

Study Analytics
Name
Mastery
Learn
Test
Matching
Spaced

No study sessions yet.

88 Terms

1
New cards

NET: Network layer availability

NET: In every host and router

2
New cards

NET: 2 key network-layer functions

NET: forwarding and routing

3
New cards

NET: Forwarding

NET: move packets from router's input to appropriate router's output

4
New cards

NET: Routing

NET: determine route take by packets from source to destination

5
New cards

NET: Why use address ranges?

NET: 4 billion IP addresses

6
New cards

NET: longest prefix matching

NET: when looking for forwarding table entry for given destination address, use longest address prefix that matching destination address

7
New cards

NET: 2 key router functions

NET: run routing protocols, forward datagrams

8
New cards

NET: routing architecture

NET: routing processor (compute forwarding tables, routing, management control plane, done in software), high-speed switching fabric (forwarding data plane, done in hardware)

9
New cards

NET: Input ports

NET: line termination (physical layer); link layer protocol to receive; lookup, forwarding, queueing

10
New cards

NET: decentralized switching

NET: given datagram destination, lookup output port using forwarding table in input port memory. Complete input port processing at line speed. Queuing if datagrams arrive faster than forward rate into switch fabric

11
New cards

NET: switching fabric

NET: transfer packet from input buffer to output buffer

12
New cards

NET: switching rate

NET: rate at which packets are transfered from input to output, desirable at N * line rate, with N inputs

13
New cards

NET: switching fabric types

NET: memory, bus, crossbar

14
New cards

NET: switching via memory

NET: switching under control of CPU, packet copied to main memory, speed limited by memory bandwidth

15
New cards

NET: switching via bus

NET: datagram moves via a shared bus, bus contention (speed limited by bus bandwidth)

16
New cards

NET: switching via crossbar

NET: overcome bus bandwidth, connect processors to multiprocessors, parallel transmissions

17
New cards

NET: output ports

NET: datagram buffer and queue, link layer protocol to send, line termination

18
New cards

NET: output port buffering

NET: required when datagrams arrive from fabric faster than transmission rate. queuing and loss due to buffer overflow.

19
New cards

NET: scheduling discipline

NET: choose among queued datagrams for transmission

20
New cards

NET: input port queue

NET: fabric slower than input ports combined, loss due to input buffer overflow

21
New cards

NET: head-of-the-line (HOL) blocking

NET: queued datagram at front of queue prevents others in queue from moving forward

22
New cards

NET: Network Layer breakdown

NET: Routing protocols (path selection, RIP, OSPF, BGP) that feeds into forwarding table; IP protocol (address conventions, datagram format, packet handling conventions); ICMP protocol (error reporting, router signaling)

23
New cards

NET: 13 Header Segments in IP datagram

NET: version, header length, type of service, length, 16-bit id, flags, fragment offset, time to live, upper layer, header checksum, source ip address, destination ip address, options

24
New cards

NET: How much overhead in ip datagram?

NET: 20 bytes of TCP + 20 bytes IP = 40 bytes + app overhead

25
New cards

NET: MTU

NET: network links have maximum transfer size - largest link-level frame, different links have different MTU

26
New cards

NET: ip fragmentation

NET: large ip datagram divided and reassembled at final destination. ip header bits (16-bit id, flag, fragment offset) used to identify and order related fragments

27
New cards

NET: ip address

NET: 32-bit identifier for host, router interface

28
New cards

NET: interface

NET: connection between host/router and physical link

29
New cards

NET: subnet

NET: device interfaces with same subnet part of ip address, can physically reach other without intervening router

30
New cards

NET: high order bits

NET: subnet part in ip address

31
New cards

NET: low order bits

NET: host part in ip address

32
New cards

NET: how to create a subnet

NET: detach each interface from its host or router, creating isolated networks

33
New cards

NET: ip address format

NET: a.b.c.d/x, x is number of bits in subnet

34
New cards

NET: how host get ip address

NET: DHCP or hardcoded by system admin in a file

35
New cards

NET: DHCP

NET: dynamic host configuration protocol

36
New cards

NET: DHCP goals

NET: dynamically get address from server, renew its lease on address in use, allows reuse of addresses, support for mobile users

37
New cards

NET: DHCP message

NET: DHCP discover, DHCP offer, DHCP request, DHCP ack

38
New cards

NET: what DHCP returns

NET: network mask, DNS name and ip address, first-hop router address

39
New cards

NET: DHCP encapsulation

NET: UDP

40
New cards

NET: how network get subnet part of ip address

NET: gets an allocation of ISP addresses; hierarchical address allows efficient advertisement of route information

41
New cards

NET: how isp get block of address

NET: ICANN

42
New cards

NET: ICANN

NET: internet corporation for assigned names and numbers

43
New cards

NET: ICANN services

NET: allocates addresses, resolve disputes, manages DNS, assigns domain names

44
New cards

NET: NAT purpose

NET: all datagrams leaving local network have same ip address

45
New cards

NET: NAT implementation

NET: 1) replace source ip and port with NAT ip and port in outgoing datagrams

2) remember translations in NAT table

3) replace NAT ip and port with destination ip and port in incoming datagrams

46
New cards

NET: NAT address count

NET: 16-bit port number -> 60,000 connections

47
New cards

NET: NAT controversy

NET: routers should process up to layer 3, violates end-to-end argument, address shortage should be solved by IPv6

48
New cards

NET: NAT traversal solutions

NET: 1) statically configure NAT to forward requests at given port

2) UPnP allows NAT to learn public IP address and add/remove port mappings

3) establish relay to bridge connections

49
New cards

NET: ICMP

NET: internet control message protocol

50
New cards

NET: ICMP definition

NET: used by hosts/routers to communicate errors (unreachable host, network, port, protocol), echo request/reply, carried in IP datagrams

51
New cards

NET: ICMP message

NET: type, code, first 8 bytes of ip datagram causing error

52
New cards

NET: Traceroute and ICMP

NET: sources sends udp segments, routers discard datagrams and send ICMP messages, source records RTTs

53
New cards

NET: traceroute stopping criteria

NET: segments reaches destination, port unreachable error, source stops

54
New cards

NET: IPv6 motivation

NET: 32-bit address space soon to be allocated, header speed processing, header facilitate QoS

55
New cards

NET: IPv6 changes

NET: fixed 40-byte header

no fragmentation

prioritizing

no checksum

flow labeling

56
New cards

NET: ICMPv6

NET: additional message types like "Packet too Big"

multicast functions

57
New cards

NET: tunneling

NET: IPv6 datagram carried as payload in IPv4 datagram among IPv4 routers

58
New cards

NET: Global routing

NET: all routers have complete graph topology and link costs (link state)

59
New cards

NET: decentralized routing

NET: router knows neighbors link costs, iterative process of computation, exchange of information with neighbors (distance vector)

60
New cards

NET: static routing

NET: routes change slowly over time

61
New cards

NET: dynamic routing

NET: routes change quickly, periodic updates in response to link cost changes

62
New cards

NET: link-state routing

NET: Dijkstra's algorithm, link costs known to all nodes, O(n^2), O(nlogn) with efficient implementations

63
New cards

NET: distance-vector routing

NET: Bellman-Ford algorithm

64
New cards

NET: Poison reverse

NET: If Z routes through Y to get to X, Y won't route through X to get to Z if Z tells Y its distance to X is infinite

65
New cards

NET: hierarchical routing purpose

NET: 600 million destination routers, can't store all destinations in routing tables, routing table exchanges will swamp links

66
New cards

NET: administrative autonomy

NET: each network admin may want to control routing in its own network

67
New cards

NET: autonomous systems

NET: (AS) aggregate routers into regions, routers in same AS run same routing protocol, forwarding table configured by intra-AS and inter-AS, intra-AS determines internal paths, intra-AS and inter-AS determines external paths

68
New cards

NET: gateway router

NET: at edge of AS, has link to router in another AS

69
New cards

NET: interior gateway protocols

NET: IGP, aka intra-AS routing protocol

70
New cards

NET: 3 intra-AS protocols

NET: RIP, OSPF, IGRP

71
New cards

NET: RIP

NET: routing information protocol

distance vector

15 max hops

16 hops = infinity

1 link cost

DV's advertised every 30 sec

advertisement list 25 destination subnets

72
New cards

NET: RIP trigger

NET: if no advertisement heard after 180 seconds, declare link dead and invalidate those routes, send new advertisements to neighbors, poison reverse prevents ping-pong loops

73
New cards

NET: RIP table management

NET: managed by application-layer route-d process (daemon), sent in UDP packets

74
New cards

NET: OSPF

NET: Open Shortest Path First

publicly available

link state

carried over IP

75
New cards

NET: IS-IS routing protocol

NET: identical to OSPF

76
New cards

NET: OSPF advanced features

NET: security - messages authenticated

multicast support

multiple cost metrics for different ToS

multiple same-cost paths

hierarchical in large domains

77
New cards

NET: hierarchical OSPF

NET: 2 levels (backbone and local area)

advertisements only in area

area border routers (summarize internal area, sends to other areas)

backbone router (run OSPF in backbone)

boundary router (connect to other AS)

78
New cards

NET: BGP

NET: border gateway protocol

most important routing protocol

uses semi-permanent TCP

only inter-AS protocol

79
New cards

NET: eBGP

NET: obtain subnet reachability information from neighbors

80
New cards

NET: iBGP

NET: propagates reachability information to its AS

81
New cards

NET: BGP attributes

NET: AS-PATH (ASs through which prefix advertisement has passed)

NEXT-HOP (specific internal-AS router to next-hop AS)

82
New cards

NET: policy-based routing

NET: gateway router receiving route advertisement uses import policy to accept/decline

83
New cards

NET: BGP route selection

NET: 1) local preference/policy decision

2) shortest AS-PATH

3) closest NEXT-HOP (hot potato)

4) additional criteria

84
New cards

NET: BGP messages

NET: OPEN (open tcp connection)

UPDATE (advertise new path)

KEEPALIVE (keep connection alive)

NOTIFICATION (report errors/close connections)

85
New cards

NET: broadcasting

NET: deliver packets from source to multiple destinations

86
New cards

NET: flooding

NET: when node receives broadcast packet and sends copy to all its neighbors

87
New cards

NET: controlled flooding

NET: node only broadcast packet if it hasn't broadcast the same packet before

88
New cards

NET: reverse path forwarding

NET: only forward packet if it arrived on shortest path between node and source