1. 1 Songwu Lu/UCLACS118/3-14-2002 Computer Networking: the overall picture  why packet switching?  What’s in a packet: –header: contains all the information needed for data delivery  layered network protocol architecture: why layering? –Divide and conquer Switch 2 Switch 1 Packet (carries destination address) header data Efficient resource sharing Flexible delivery

2. 2 Songwu Lu/UCLACS118/3-14-2002 On Final Exam  TAs will give additional review lectures  Office hours next week: –Monday, 1-2pm –Tuesday, 1-3pm –Thursday, 1-3pm  For your current standing in the class, check with your TA  You can bring two (8x11) sheets into the final exam

3. 3 Songwu Lu/UCLACS118/3-14-2002 Materials to be covered  Main focus (about 80%): –Chapters 4 and 5  Additional chapters (about 20%): –Chapter 6 –Chapter 3

4. 4 Songwu Lu/UCLACS118/3-14-2002 Exam Format  Multiple choices: –Which of the following error detection algorithms can guarantee 100% error detection? –(A)CRC; (B) Internet Checksum; © both CRC and checksum; (D) two-dimensional parity; (E) none of the above  Short Q & A: –Does a host know the local DNS server by its DNS name? Or by its IP address? Or by its Ethernet address? Explain

5. 5 Songwu Lu/UCLACS118/3-14-2002 Exam Format  Standard Problem solving –Given a network topology, compute the minimum cost path from the source to the destination by applying the distance vector routing algorithm

6. 6 Songwu Lu/UCLACS118/3-14-2002 Computer Networking: the overall picture  why packet switching?  What’s in a packet: –header: contains all the information needed for data delivery  layered network protocol architecture: why layering? –Divide and conquer Switch 2 Switch 1 Packet (carries destination address) header data Efficient resource sharing Flexible delivery

7. 7 Songwu Lu/UCLACS118/3-14-2002 What are the layers?  Network layer: transmit packets from one host to another host in the Internet –issues: routing, switching, multicast  Internet Protocol (IP):deliver packets host  host –IP packet format –packet fragmentation and reassembly –delivery by encapsulation –IP addresses: class-based addresses, subnetting, CIDR

8. 8 Songwu Lu/UCLACS118/3-14-2002 More protocol layers  link layer: send data frames between directly connected nodes –framing –bit error detection (parity check, checksum, CRC) –media-access-control (MAC): –Ethernet, token ring –Hubs, bridges, and switches  Higher layers above network layer –transport protocols:reside in end hosts only –application protocols

9. 9 Songwu Lu/UCLACS118/3-14-2002 Ethernet frame IP packet TCP packet header tail IP hdr TCP hdr DATA Layered protocol implementation A protocol defines: u the format of message exchanged between peer entities u the actions taken on receipt of the message What’s in the header: all the information,and only the information that’s needed for the protocol’s functionality

10. 10 Songwu Lu/UCLACS118/3-14-2002 Network layer design  Network service model  Routing  IPv4 and IPv6  Multicast routing

11. 11 Songwu Lu/UCLACS118/3-14-2002 Network Service Model  Datagram vs. virtual circuit –What is the difference? –What kind of service model the Internet is using? –What are the cons and pros of Internet service model? –Is it appropriate to support data / multimedia?  How can we realize connection-oriented services and connection-less services? –At the network layer –At the transport layer

12. 12 Songwu Lu/UCLACS118/3-14-2002 Network routing  routing protocol: distributed way to compute shortest path to all destinations  distance-vector routing protocol –A node’s update includes a list of [destination, distance] pairs for all destinations that A knows of –send routing updates to neighbor nodes only  link-state routing protocol –A node’s update includes a list of [neighbor, link distance] pairs for all its links –routing updates are flooded to the entire network

13. 13 Songwu Lu/UCLACS118/3-14-2002 What you need to know about routing algorithms?  Given a topology with link cost, how to compute the minimum cost path from a given source to a given destination ? –apply distance vector routing algorithm –apply link state routing algorithm  On distance vector routing: –What does it mean “good news travels fast”? –What is the “count-to-infinity” problem? –What are the fixes to count-to-infinity problem? Can they solve the problem completely?  Comparisons between DV and LS routing: –Message complexity, convergence speed, robustness

14. 14 Songwu Lu/UCLACS118/3-14-2002 More on routing  On virtual circuit routing: –What is shortest path first routing? –What is least loaded path routing? –What is maximum free circuit routing?  Why do we need hierarchical routing over the Internet? –two reasons:

15. 15 Songwu Lu/UCLACS118/3-14-2002 An example for Distance Vector routing initial state at each node A G H D F 1 2 3 2 4 1 1 2 3 4 4 B 1 B C ? D ? E ? F ? G ? Dst Dis Nex H 2 H A's routing table A 1 A C 2 C D ? E ? F ? G ? Dst Dis Nex H 3 H B's routing table A ? B 2 B D 1 D E 1 E F ? G ? Dst Dis Nex H ? C's routing table A ? B ? C 1 C E ? F 4 F G ? Dst Dis Nex H ? D's routing table B C A ? B 2 B C 1 C D ? F 3 F G 2 G H ? E's routing table E

16. 16 Songwu Lu/UCLACS118/3-14-2002 Routing table after one update B 1 B C 3 B D ? E ? F ? G ? H 2 H A's routing table after receiving an update from B B's routing table after receiving an update from C A 1 A C 2 C D 3 C E 3 C F ? G ? H 3 H B 1 B C ? D ? E ? F ? G ? Dst Dis Nex H 2 H A's routing table A 1 A C 2 C D ? E ? F ? G ? Dst Dis Nex H 3 H B's routing table A ? B 2 B D 1 D E 1 E F ? G ? Dst Dis Nex H ? C's routing table A ? B 2 B C 1 C D ? F 3 F G 2 G H ? E's routing table B 3 B D 1 D E 1 E F 4 E G 3 E H 2 H C's routing table after receiving an update from E A ?

17. 17 Songwu Lu/UCLACS118/3-14-2002 An example for link state protocol: A's topology table A G H D F 1 2 3 2 4 1 1 2 3 4 4 B C E 0 1 2 0 0 0 0 0 A B C D E F G H 0 ABCDEFGHABCDEFGH 0 0 1 3 1 0 2 3 0 0 0 0 A B C D E F G H 0 ABCDEFGHABCDEFGH 0 After update from B 0 1 2 1 0 2 3 0 0 1 0 3 2 0 A B C D E F G H 0 ABCDEFGHABCDEFGH 0 After update from E

18. 18 Songwu Lu/UCLACS118/3-14-2002 A G H F 1 2 3 2 4 1 1 2 3 4 4 B C E 0 1 2 1 0 2 3 2 0 1 1 1 0 4 1 0 3 2 4 3 0 4 A B C D E F G H 2 4 0 4 ABCDEFGHABCDEFGH 2 3 4 0 After A has received an update from every other node D

19. 19 Songwu Lu/UCLACS118/3-14-2002 An example for Dijkstra Algorithm G H F 1 2 3 2 4 1 1 2 3 4 4 B C E D A (0) {A}, AB=1, AC= , AD= , AE= , AF= , AG= , AH=2 (2) {A, B, H}, Nh=H, update: AC=3, AD= , AE= , AF= , AG=AH+HG=6 (1) {A, B}, Nb= B, update: AC=AB+BC=3, AD= , AE= , AF= , AG= , AH=2 (3) {A, B, H, C}, Nc=B, update: AD=AC+CD=4, AE=AC+CE=4, AF= , AG=6 (4) {A, B, H, C, D}, Nd=B, update: AE=4, AF=AD+DF=8, AG=6 (5) {A, B, H, C, D, E}, Ne=B, update: AF=AE+EF=7, AG=6 (6) {A, B, H, C, D, E, G}, Ng=B, update: AF=7 (7) {A, B, H, C, D, E, G, F} Nf=B 0 1 3 4 4 7 6 2 A B C D E F G H A

20. 20 Songwu Lu/UCLACS118/3-14-2002 Internet Protocol (IP)  Relation between IP & network layer –IP is responsible for host -to-host packet delivery, normally through a chain of IP routers –physical networks do the real work of getting packets from one IP node to the next  understand IPv4 addressing –IP address structure »two-level hierarchy Network ID + host ID »class-based address: Class A, B, C, D »subnetting

21. 21 Songwu Lu/UCLACS118/3-14-2002 CIDR: Classless InterDomain Routing  assign network addresses by blocks of contiguous IP addresses, in a form of –mask identifies block size, must be power of 2 –example: allocation of 4 class-C address blocks 192.4.16.0  192.4.19.255,, or 192.4.16/22 11000000000001000001000000000000 11000000000001000001001111111111 192 4 16 0 192 4 19 255 11111111111111111111110000000000 255 255 252 0

22. 22 Songwu Lu/UCLACS118/3-14-2002 Forwarding IP Packets  hosts decide whether the destination is on the same network, if not, send packet to a default router  routing: IP router looks up the forwarding table to determine the next hop to forward the packet to –routers may also use a default router for far- away destinations  address translation: mapping an IP address to physical network address: ARP  packet encapsulation and decapsulation when crossing each physical network

23. 23 Songwu Lu/UCLACS118/3-14-2002 IPv4 Header  What is the purpose for each header field? –For example, why do we need TTL? Why do we need Options?  IP Fragmentation and reassembly: –Given a large IP packet passing through a small- packet network, how does fragmentation work? The reassembly process? –Where do we reassembly the fragments? Why?  On ICMP: how does traceroute work?

24. 24 Songwu Lu/UCLACS118/3-14-2002 Internet routing  What is the main difference between RIP and BGP?  How do we further reduce the routing table size? –Default route

25. 25 Songwu Lu/UCLACS118/3-14-2002 IPv6  What header fields are available in both IPv4 and IPv6?  What are the new fields in IPv6 header? Why do we need them?  Comparison btw IPv4 and IPv6  Does IPv6 allow fragmentation or not? If not, how does IPv6 handle the case of small-packet-network?

26. 26 Songwu Lu/UCLACS118/3-14-2002 Transition from IPv4 to IPv6  What is dual-stack solution?  On tunneling: how does tunneling work? What is the advantage? Incrementally deploy new protocols: IPv6, IP multicast, etc….

27. 27 Songwu Lu/UCLACS118/3-14-2002 IP Multicasting  IP multicast service model –each group identified by an IP mcast address –members can be anywhere –members may join and leave any time  How to map IP mcast addr to a link-layer mcast addr (not in the exam) –place the low-order 23 bits of IP mcast addr to the lower 23 bits of Ethernet mcast addr: 01.00.5E.00.00.00  IGMP protocol –hosts report group membership to a local router

28. 28 Songwu Lu/UCLACS118/3-14-2002 IGMP  A query router in each link  querier periodically polls the link  on receipt of query message, hosts set a random timer for each mcast group it belongs to  when timer expires, send a report to group G  others in G hear the report and stop the timers  joining: send reports immediately  leaving: send a leave msg, and querier sends group-specific queries again

29. 29 Songwu Lu/UCLACS118/3-14-2002 More on IGMP  Why do we need a timer in IGMP?  Does IGMP know how many active receivers in each multicast group? Why?  How does IGMP handle: –An existing member’s departure –A new member join –A failing receiver

30. 30 Songwu Lu/UCLACS118/3-14-2002 Multicast routing  distance vector mcast routing (DVMRP) –reverse path broadcast: only broadcast over output interfaces if the input interface is on the shortest path to the source S »You have to know this ! –pruning and grafting –Given a topology, know how to apply RPF to solve the problem  Mbone: –use IP tunneling to connect mcast routers

31. 31 Songwu Lu/UCLACS118/3-14-2002 Link layer design  Error detection  Framing: byte stuffing  MAC protocols  Bridges, hubs

32. 32 Songwu Lu/UCLACS118/3-14-2002 Error detection  Parity bit, 2D parity bit –Can they detect all error cases?  CRC –What is the main idea of CRC? –What is the advantage to use CRC?

33. 33 Songwu Lu/UCLACS118/3-14-2002 MAC protocols  What are the three MAC protocol families?  What protocols to use for the following scenarios? – wired voice, wireless data, ftp over wired network  Compare TDMA and token-based protocol  Compare Aloha, slotted Aloha, and CSMA/CD

34. 34 Songwu Lu/UCLACS118/3-14-2002 More on MAC  CSMA/CD: –What is carrier sensing? –How to do collision detection? –Can carrier sensing avoid collisions completely? Why?  Wireless MAC (802.11) –What is hidden terminal problem? –Why do we need RTS-CTS? –Why do we need ACK?  Taking turns –How can we design a protocol to provide the best of both worlds: channel partitioning and random access?

35. 35 Songwu Lu/UCLACS118/3-14-2002 Ethernet Address and ARP  Compare ethernet address with IP address  The detailed operations of ARP  If the underlying network does not have a broadcast medium, can we still use ARP?  Design a solution so that ARP can work in point-to-point link scenarios.  The steps to send a datagram to a node off the LAN

36. 36 Songwu Lu/UCLACS118/3-14-2002 Ethernet  Ethernet MAC: –What is BEB? Why use it? –What is the capture effect?

37. 37 Songwu Lu/UCLACS118/3-14-2002 Connecting LANs  Compare hubs, bridges and routers  The exact operations of bridge learning algorithm (bridge filtering and forwarding)  Why do we need to build spanning trees? What failure happens, what to do?  Wireless Ethernet MAC: –What is hidden terminal problem? –Does wireless Ethernet MAC implement collision detection? Why? –The steps in 802.11 MAC, how RTS-CTS helps?

38. 38 Songwu Lu/UCLACS118/3-14-2002 Multimedia networking  What is the new requirement to support multimedia?  On client-side buffering: –How does it solve the jitter problem?  How do the FEC and the interleaving work?  What is the difference between guaranteed service and controlled load service?

39. 39 Songwu Lu/UCLACS118/3-14-2002 ON TCP  How does TCP handle reliable transfer? –What are the fundamental mechanisms? –How do they work?  How does TCP perform flow control?  How does TCP perform congestion control?  How does TCP estimate its RTT and retransmission timeout (RTO)?

40. 40 Songwu Lu/UCLACS118/3-14-2002 What you can do in the future?  Networking will be (already is in some sense) the next big wave of computer and information technology  It is just the end of the beginning  merging key industry sectors: PC & handheld devices, communications, and even system control (automobile etc.)  It is time for you to think about making your personal landmark for the emerging technology  How to do it: doing the right thing at the right time in a right way