Outline • Methods for packet forwarding 15-441 Computer Networking • Traditional IP addressing Lecture 7 – IP Addressing and Forwarding • CIDR IP addressing • Forwarding example Lecture 7: 9 -18-01 2 Techniques for Forwarding Packets Source Routing • Source routing • List entire path in packet • Packet carries path • Driving directions (north 3 hops, east, etc..) • Table of virtual circuits • Router processing • Connection routed through network to setup • Examine first step in directions state • Packets forwarded using connection state • Strip first step from packet • Table of global addresses (IP) • Forward to step just stripped off • Routers keep next hop for destination • Packets carry destination address Lecture 7: 9 -18-01 3 Lecture 7: 9 -18-01 4 Source Routing Example Source Routing • Advantages • Switches can be very simple and fast Packet 3,4,3 4,3 • Disadvantages 2 2 Sender R1 R2 1 3 1 3 • Variable (unbounded) header size 4 4 3 • Sources must know or discover topology (e.g., 2 failures) 1 R3 Receiver 3 • Typical use 4 • Ad-hoc networks (DSR) • Machine room networks (Myrinet) Lecture 7: 9 -18-01 5 Lecture 7: 9 -18-01 6 1 Global Addresses (IP) Global Address Example • Each packet has destination address • Each switch has forwarding table of Packet R R destination à next hop 2 2 Sender R1 R2 R à 4 1 3 1 3 • At v and x: destination à east 4 4 • At w and y: destination à south R à 3 R 2 1 • At z: destination à north R3 Receiver 3 4 R • Distributed routing algorithm for calculating R à 3 forwarding tables Lecture 7: 9 -18-01 7 Lecture 7: 9 -18-01 8 Router Table Size Global Addresses • One entry for every host on the Internet • Advantages • 100M entries,doubling every year • Stateless – simple error recovery • One entry for every LAN • Disadvantages • Every host on LAN shares prefix • Every switch knows about every destination • Still too many, doubling every year • Potentially large tables • One entry for every organization • All packets to destination take same route • Every host in organization shares prefix • Requires careful address allocation Lecture 7: 9 -18-01 9 Lecture 7: 9 -18-01 10 How do we set up Routing Tables? Virtual Circuits/Tag Switching • Graph theory to compute “shortest path” • Connection setup phase • Switches = nodes • Use other means to route setup request • Each router allocates flow ID on local link • Links = edges • Creates mapping of inbound flow ID/port to outbound • Delay, hops = cost flow ID/port • Need to adapt to changes in topology • Each packet carries connection ID • Sent from source with 1st hop connection ID • Router processing • Lookup flow ID – simple table lookup • Replace flow ID with outgoing flow ID • Forward to output port Lecture 7: 9 -18-01 11 Lecture 7: 9 -18-01 12 2 Virtual Circuits Examples Virtual Circuits • Advantages • More efficient lookup (simple table lookup) • More flexible (different path for each flow) Packet 5 7 2 2 • Can reserve bandwidth at connection setup Sender R1 R2 1,7 à 4,2 1 3 1 3 • Easier for hardware implementations 4 4 1,5 à 3,7 2 • Disadvantages 2 • Still need to route connection setup request 1 R3 Receiver 3 • More complex failure recovery – must recreate 4 6 2,2 à 3,6 connection state • Typical uses • ATM – combined with fix sized cells • MPLS – tag switching for IP networks Lecture 7: 9 -18-01 13 Lecture 7: 9 -18-01 14 IP Datagrams on Virtual Circuits IP Datagrams on Virtual Circuits • Challenge – when to setup connections • Traffic pattern • At bootup time – permanent virtual circuits • Few long lived flows (PVC) • Flow – set of data packets from source to • Large number of circuits destination • For every packet transmission • Large percentage of packet traffic • Connection setup is expensive • Improving forwarding performance by using • For every connection virtual circuits for these flows • What is a connection? • Other traffic uses normal IP forwarding • How to route connectionless traffic? Lecture 7: 9 -18-01 15 Lecture 7: 9 -18-01 16 Comparison Forwarding vs. Routing • Forwarding: the process of moving packets from Source Routing Global Addresses Virtual Circuits input to output Header Size Worst OK – Large address Best • The forwarding table Number of hosts • Information in the packet Router Table Size None Number of circuits (prefixes) • Routing: process by which the forwarding table is Forward Overhead Best Prefix matching Pretty Good built and maintained Setup Overhead None None Connection Setup • One or more routing protocols • Procedures (algorithms) to convert routing info to Tell all routers and Error Recovery Tell all hosts Tell all routers Tear down circuit forwarding table. and re-route Lecture 7: 9 -18-01 17 Lecture 7: 9 -18-01 18 3 Outline How is IP Design Standardized? • Methods for packet forwarding • IETF • Voluntary organization • Meeting every 4 months • Traditional IP addressing • Working groups and email discussions • “We reject kings, presidents, and voting; we believe in rough consensus and running code” • CIDR IP addressing (Dave Clark 1992) • Need 2 independent, interoperable implementations for standard • Forwarding example • IRTF • End2End • Reliable Multicast, etc.. Lecture 7: 9 -18-01 19 Lecture 7: 9 -18-01 20 Addressing in IP Addressing Considerations • IP addresses are names of interfaces • Fixed length or variable length? • Domain Name System (DNS) names are • Issues: names of hosts • Flexibility • DNS binds host names to interfaces • Processing costs • Routing binds interface names to paths • Header size • Engineering choice: IP uses fixed length addresses Lecture 7: 9 -18-01 21 Lecture 7: 9 -18-01 22 Addressing Considerations IP Addresses • Structured vs flat • Fixed length: 32 bits • Initial classful structure (1981) • Issues • Total IP address size: 4 billion • What information would routers need to route to • Class A: 128 networks, 16M hosts • Class B: 16K networks, 64K hosts Ethernet addresses? • Class C: 2M networks, 256 hosts • Need structure for designing scalable binding from interface name to route! High Order Bits Format Class • How many levels? Fixed? Variable? 0 7 bits of net, 24 bits of host A 10 14 bits of net, 16 bits of host B 110 21 bits of net, 8 bits of host C Lecture 7: 9 -18-01 23 Lecture 7: 9 -18-01 24 4 IP Address Classes (Some are Obsolete) Some Special IP Addresses • 127.0.0.1: local host (a.k.a. the loopback Network ID Host ID 8 16 24 32 address Class A 0 Network ID Host ID • Host bits all set to 0: network address Class B 10 • Host bits all set to 1: broadcast address Class C 110 Class D 1110 Multicast Addresses Class E 1111 Reserved for experiments Lecture 7: 9 -18-01 25 Lecture 7: 9 -18-01 26 Interaction with Link Layer Original IP Route Lookup • How does one find the Ethernet address of • Address classes a IP host? • A: 0 | 7 bit network | 24 bit host (16M each) • ARP • B: 10 | 14 bit network | 16 bit host (64K) • Broadcast search for IP address • C: 110 | 21 bit network | 8 bit host (255) • E.g., “who-has 128.2.184.45 tell 128.2.206.138” sent to Ethernet broadcast (all FF address) • Address would specify prefix for forwarding • Destination responds (only to requester using table unicast) with appropriate 48-bit Ethernet • Simple lookup address • E.g, “reply 128.2.184.45 is-at 0:d0:bc:f2:18:58” sent to 0:c0:4f:d:ed:c6 Lecture 7: 9 -18-01 27 Lecture 7: 9 -18-01 28 Original IP Route Lookup – Subnet Addressing Example RFC917 (1984) • www.cmu.edu address 128.2.11.43 • For class B & C networks • Class B address – class + network is 128.2 • Very few LANs have close to 64K hosts • Lookup 128.2 in forwarding table • For electrical/LAN limitations, performance or administrative reasons • Prefix – part of address that really matters for routing • Need simple way to get multiple “networks” • Forwarding table contains • Use bridging, multiple IP networks or split up single • List of class+network entries network address ranges (subnet) • A few fixed prefix lengths (8/16/24) • Must reduce the total number of network addresses that are assigned • Large tables • CMU case study in RFC • 2 Million class C networks • Chose not to adopt – concern that it would not be widely supported J Lecture 7: 9 -18-01 29 Lecture 7: 9 -18-01 30 5 Subnetting Subnetting Example • Variable length subnet masks • Assume an organization was assigned • Could subnet a class B into several chunks address 150.100 • Assume < 100 hosts per subnet • How many host bits do we need? Network Host • Seven Network Subnet Host • What is the network mask? • 11111111 11111111 11111111 10000000 1111.. ..1111 00000000 Mask • 255.255.255.128 Lecture 7: 9 -18-01 31 Lecture 7: 9 -18-01 32 Forwarding Example Outline • Assume a packet arrives with address • Methods for packet forwarding 150.100.12.176 • Step 1: AND address with class +subnet mask 150.100.12.154 150.100.12.176 • Traditional IP addressing H1 H2 150.100.12.128 • CIDR IP addressing 150.100.12.129 150.100.12.24 150.100.12.55 150.100.0.1 R1 H3 H4 To Internet • Forwarding example 150.100.12.4 150.100.12.0 Lecture 7: 9 -18-01 33 Lecture 7: 9 -18-01 34 IP Address Problem (1991) IP Address Utilization (‘98) • Address space depletion • In danger of running out of classes A and B • Why? • Class C too small for most domains • Very few class A – very careful about giving them out • Class B – greatest problem http://www.caida.org/outreach/resources/learn/ipv4space/ Lecture 7: 9 -18-01 35 Lecture 7: 9 -18-01 36 6 IP Addressing Problems Classless Inter-Domain Routing • Class B sparsely populated • Do not use classes to determine network ID • But people refuse to give it back • Assign any range of addresses to network • Large forwarding tables • Use common part of address as network number • 2 Million possible class C groups • E.g., addresses 192.4.16 - 192.4.31 have the first 20 • Solution bits in common.