Where are we?

1. Where are we? • Chapter 3 and 4 are focused on getting the data from one place to another. • Switching and routing • Review the next slides • First our goal is to have applications on two hosts communicating with each other • The data needs to be augmented so it can be sent to the destination, understood and redirected at all points in between and interpreted correctly when it arrives.

2. Host Host Application Host Channel Application Host Host

3. Application programs Process-to-process channels Host-to-host connectivity Hardware

4. Encapsulation • Notice how the data is encapsulated in a HHP header that has addressing information and the RRP header that tells the destination which application to apply to the data.

5. Host 1 Host 2 Application Application program program Data Data RRP RRP RRP Data RRP Data HHP HHP HHP RRP Data

6. Chapter 3: Packet Switching • Core job of switches – take packets that arrive on an input port and forward (or switch) them to the correct output port • Key problem – finite bandwidth of outputs, packets coming from several sources that all need to be switched to the same destination port can overload the capacity (congestion) of the output port.

7. Chapter Outline • Types of Switching • Datagrams • forwarding tables • Virtual circuits • Source routing • Bridges and LAN switches • Learning bridges • Spanning tree algorithm • Broadcast and multicast

8. Chapter Outline - continued • ATM • Cells • Segmentation and reassembly • Switching hardware • Throughput • Scalability • Ports and fabric

10. Switching protocol T3 T3 STS-1

11. T3 T3 Switch T3 T3 STS-1 STS-1 Input Output ports ports

12. Switching and Forwarding • Different types of media for inputs and outputs is common • Basic idea (figure 3.3) – packet comes in on input port, switch forwards it to the correct output port • Basic question: how does the switch decide which output port to place each packet on? • Three approaches

13. Three approaches • Datagrams • Enough info included with each datagram to allow any switch along the way to decide where to send the datagram. This approach uses forwarding tables • Virtual circuits • A Virtual Circuit Identifier (VCI) is carried along with each packet, this information tells each switch what path (circuit) to forward it on.

14. continued • Source routing • Put a list of ports in the path from the source to the destination in each packet. • Note: this is different then the VCI approach, which gives each packet a VCI and that along with the input port will tell the next switch in line what to do with the packet

15. Datagrams • Study the next figure and the forwarding table on page 175. • Also the ICND book has a better, indepth discussion of forwarding tables and learning

16. Host D Host E 0 Switch 1 Host F 3 1 Switch 2 2 Host C 2 3 1 0 Host A 0 Switch 3 Host B Host G 1 3 2 Host H

17. Virtual Circuit • Look over the next figure • 2nd paragraph on page 177 details how this all works

18. 0 Switch 1 3 1 2 Switch 2 2 3 1 5 11 0 Host A 7 0 Switch 3 1 3 4 Host B 2

19. 8 16 8 Variable 16 8 Flag Frame Flag Address Control Data (0x7E) checksum (0x7E)

20. Source Routing • Look over figure 3.7 (next)

22. Header entering D C B A D C B A Ptr D C B A switch Header leaving A D C B D C B Ptr D C B A switch (a) (b) (c)

23. Workstation as Packet Switch • You don’t have to have specialized hardware to do switching (or routing from Chapter 4). • A workstation can do it as shown in figure 3.9 • Performance is limited however

25. Bridges and LAN Switches • Learning bridges • See ICND for a more detailed discussion of this (important) concept • Changes in hosts and switches must propagate through the network of switches • Loops are major problem • Spanning tree algorithm • Again a good discussion in ICND • Know terminology

26. A B C Port 1 Bridge Port 2 X Y Z

27. A B B3 C B5 D B7 K B2 E F B1 G H B6 B4 I J

30. Limitations of Bridges • Scalability • Should only connect a few (tens) LANs with bridges • No really good way to impose hierarchy on an extended set of LANs with bridges • However VLANs can help • Bridges forward all broadcasts (routers don’t)

31. W X VLAN 100 VLAN 100 B1 B2 VLAN 200 VLAN 200 Y Z

32. ATM • Will not spend as much time on this as Ethernet • Connection-oriented, packet-switched technology • Cells • Fixed cell size • See page 199 (points 1 and 2) for key advantages • Format – see next figure

33. 4 16 3 1 8 8 384 (48 bytes) GFC VPI VCI Type CLP HEC (CRC-8) Payload

34. Segmentation and reassembly • 48 byte ATM packets don’t hold much information • Larger messages must be fragmented and the put back together at the destination • ATM Adaptation Layer (AAL) is a protocol layer added that sits between ATM and a variable-length protocol that might use ATM like IP

35. AAL AAL … … A TM A TM

36. CS-PDU CS-PDU User data header trailer 44 bytes 44 bytes 44 bytes 44 bytes AAL header AAL trailer ATM header Cell payload Padding

38. ATM in the LAN • Section 3.3.5 goes over several issues when using ATM in the LAN • Switched technology vs. the shared media technology of Ethernet • Common protocols like ARP (Address Resolution Protocol) don’t work the same at all.

40. Higher-layer Higher-layer protocols protocols (IP, ARP, . . .) (IP, ARP, . . .) Ethernet-like interface Signalling Signalling + LANE + LANE AAL5 AAL5 ATM ATM ATM PHY PHY PHY PHY Host Switch Host

41. Switching Hardware • Throughput • Very difficult to define the throughput of a switch • Size/scalability • How fast does the cost of a size n switch increase as a function of n2 or n3 ? • cost

42. Hardware • Ports • Fabric • Study the knockout switch structure

43. Input Output port port Output Input port port Fabric Output Input port port Output Input port port

44. Inputs D D D D D D D D D D D D D D 1 2 3 4 Outputs

45. Shifter (a) Buffers Shifter (b) Buffers Shifter (c) Buffers

46. Shared-Media switch • See figure 3.31 • Recall the example of a workstation as a switch

47. Inputs Outputs … … Mux Buffer memory Demux W rite Read control control