1. Network Layer – part 31 Customer-Provider Routing Relationships  The Global Internet consists of Autonomous Systems (AS) interconnected with each other:  Customer: Stub AS: small corporation  Customer: Multihomed AS: large corporation (no transit)  Provider: Transit AS: backbone provider networks A B C w x y e.g. A, B, C e.g. x e.g. w, y Advertises to its neighbors that it has no paths to any other destinations except itself All traffic entering must be destined for w, all traffic leaving must have originated from w Stub AS must be prevented from forwarding traffic between Transit ASs using Selective Route Advertisement Policy Group of routers

2. Network Layer – part 32 Routing in the Internet  Two-level routing:  Intra-AS: administrator is responsible for choice  Inter-AS: unique standard Border Gateway Protocol (BGP4)  de facto standard inter-AS routing protocol in today’s Internet  provides each AS a means to: obtain subnet reachability information (i.e. via one of its neighboring AS) propagate the reachability information to all routers internal to the AS determine “good” routes to subnets based on the reachability information and on AS policy. Allows each subnet to advertise its existence to the rest of the Internet

3. Network Layer – part 33 Internet AS Hierarchy AS border (exterior gateway) routers AS interior (gateway) routers

4. Network Layer – part 34 Intra-AS Routing  Also known as Interior Gateway Protocols (IGP)  Most common IGPs:  RIP(lower-tier ISPs and Enterprise networks)  RIP: Routing Information Protocol (lower-tier ISPs and Enterprise networks)  OSPF(upper-tier ISPs)  OSPF: Open Shortest Path First (upper-tier ISPs)  IGRP  IGRP: Interior Gateway Routing Protocol (Cisco proprietary)

5. Network Layer – part 35 RIP ( Routing Information Protocol)  Distance vector algorithm  Included in (Berkeley Software Distribution) BSD-UNIX Distribution in 1982  Distance metric: max = 15 hops  # of hops (max = 15 hops) = (AS < 15 hops in diameter)  Can you guess why? every 30 sec  Distance vectors: exchange routing updates via Response Message (also called advertisement) every 30 sec 25 destination subnets  Each advertisement: route to up to 25 destination subnets within the AS, including the sender’s distance from each of them Hop – no. of subnets traversed along the shortest path from Source Router to Destination Subnet, including the Destination Subnet.

6. Network Layer – part 36 RIP (Routing Information Protocol) Destination Subnet Next Router Num. of hops to dest. w A2 y B2 z B7 x --1.......... w xy z A C D B Routing table in Router D … Example subnet

7. Network Layer – part 37 RIP (Routing Information Protocol) Destination Subnet Next Router Num. of hops to dest. wA2 yB2 zB7 x--1.......... w xy z A C D B Routing table in Router D … Example Destination Subnet Next Router Num. of hops to dest. zC4 w-- 1 x--1.......... (30 secs. later.. D receives an advertisement from Router A ) Router A has a shorter path to Z!

8. Network Layer – part 38 RIP (Routing Information Protocol) Destination Subnet Next Router Num. of hops to dest. wA2 yB2 A zA4 x--1.......... w xy z A C D B Routing table in Router D … Example Destination Subnet Next Router Num. of hops to dest. zC4 w-- 1 x--1.......... Advertisement from Router A Router D updates its entry for destination Z

9. Network Layer – part 39 RIP: Link Failure and Recovery after 180 sec If no advertisement is heard after 180 sec --> the neighbour/link is declared dead  Modifies routing table - routes via neighbor invalidated  new advertisements sent to neighbors  neighbours in turn send out new advertisements (if tables changed)  link failure info quickly propagates to entire net  poisoned reverse used to prevent ping-pong loops (infinite distance = 16 hops) Example

10. Network Layer – part 310 Routing Info Protocol (RIP) Table processing application-level route-d (daemon)  RIP routing tables managed by application-level process called route-d (daemon) UDP packets  advertisements sent in UDP packets, periodically repeated Able to manipulate routing tables within the UNIX kernel via UDP, port 520

11. Network Layer – part 311 OSPF (Open Shortest Path First)  “Open” means publicly available Link-State algorithm  Uses Link-State algorithm  LS packet dissemination  Topology map at each node  Route computation using Dijkstra's algorithm  OSPF advertisement carries one entry per neighbor router  Advertisements disseminated to entire AS (via flooding)  Carried in OSPF messages directly over IP (rather than TCP or UDP with upper-layer protocol of 89 all Broadcasts information to all not just neighboring routers OSPF Protocol Functionalities: reliable data transfer, link-state broadcast, check for links operability, extraction of neighboring router’s database of network-wide link state

12. Network Layer – part 312 OSPF advanced features (not in RIP)  Security: all OSPF messages authenticated (to prevent malicious intrusion)  Multiple same-cost paths allowed (only one path in RIP)  Integrated uni- and multicast routing support:  Multicast OSPF (MOSPF) uses same topology data base as OSPF  Hierarchical OSPF in large domains. Allow only trusted routers Most significant advancement! Has the ability to structure an autonomous system hierarchically

13. Network Layer – part 313 Hierarchical Open Shortest Path First (OSPF)

14. Network Layer – part 314 Hierarchical OSPF  Two-level hierarchy: local area, backbone.  Link-state advertisements are sent only within an area  each node has detailed area topology; only know direction (shortest path) to nets in other areas.  Each area runs its own OSPF link-state routing algorithm  Area border routers: responsible for routing packets outside the area.  Backbone routers: run OSPF routing limited to backbone.  Boundary routers: connect to other ASs.

15. Network Layer – part 315 IGRP (Interior Gateway Routing Protocol)  CISCO proprietary; successor of RIP (mid 80s)  Uses the Distance Vector algorithm, like RIP  several cost metrics (delay, bandwidth, reliability, load, etc.)  uses TCP to exchange routing updates  Loop-free routing via Distributed Updating Alg. (DUAL) based on diffused computation

16. Network Layer – part 316 Router Architecture Overview Two key router functions:  run routing algorithms/protocol (RIP, OSPF, BGP)  switching datagrams from incoming to outgoing link Physical layer functions Data link layer functions Lookup & forwarding functions computes routing tables, performs Network management functions

17. Network Layer – part 317 Input Port Functions Decentralized switching:  given datagram dest., lookup output port using routing table in input port memory  goal  goal: complete input port processing at 'line speed'  queuing  queuing: happens if datagrams arrive faster than forwarding rate into switch fabric Physical layer: bit-level reception Data link layer: e.g., Ethernet see chapter 5

18. Network Layer – part 318 Input Port Queuing  Fabric slower than input ports combined -> queueing may occur at input queues  Head-of-the-Line (HOL) blocking: queued datagram at front of queue prevents others in queue from moving forward  queueing delay and loss due to input buffer overflow! Slot for Green packet is free, but there is HOL blocking, so Green packet will have to wait

19. Network Layer – part 319 Three types of switching fabrics 1 packet No routing processor; 1 packet at a time Like shared memory multiprocessors 2n n 2n buses that connect n input ports to n output ports

20. Network Layer – part 320 Switching Via Memory First generation routers: CPU  packet copied by system's (single) CPU memory bandwidth  speed limited by memory bandwidth (2 bus crossings per datagram) Input Port Output Port Workstation’s Memory System Bus Modern routers: processor memory  input port processor performs lookup, copy into memory  Cisco Catalyst 8500

21. Network Layer – part 321 Switching Via Bus  datagram from input port memory shared bus to output port memory via a shared bus  bus contention: switching speed limited by bus bandwidth  1 Gbps bus, Cisco 1900: sufficient speed for access and enterprise routers (not regional or backbone)

22. Network Layer – part 322 Switching Via An Interconnection Network  overcome bus bandwidth limitations  Banyan networks, other interconnection nets initially developed to connect processors in multiprocessor fragmenting datagram  Other Advanced design: fragmenting datagram into fixed length cells, switch cells through the fabric.  Cisco 12000: switches 60 Gbps through the interconnection network

23. Network Layer – part 323 Output Ports  Buffering required when datagrams arrive from the fabric faster than the transmission rate  Scheduling discipline chooses among queued datagrams for transmission

24. Network Layer – part 324 Output port queueing  buffering when arrival rate via switch exceeeds ouput line speed  queueing (delay) and loss due to output port buffer overflow! It is more advantageous to mark a packet before the buffer is full in order to provide a congestion signal to the sender

25. Network Layer – part 325 END OF SESSION

26. Network Layer – part 326 IPv6  Initial motivation: 32-bit address space completely allocated by 2008.  Additional motivation:  header format helps speed processing/forwarding  header changes to facilitate QoS  new anycast address: route to best of several replicated servers  IPv6 datagram format:  fixed-length 40 byte header  no fragmentation allowed

27. Network Layer – part 327 IPv6 Header (Cont) Priority: identify priority among datagrams in flow Flow Label: identify datagrams in same flow. (concept of flow not well defined). Next header: identify upper layer protocol for data

28. Network Layer – part 328 Other Changes from IPv4  Checksum: removed entirely to reduce processing time at each hop  Options: allowed, but outside of header, indicated by Next Header field  ICMPv6: new version of ICMP  additional message types, e.g. ''Packet Too Big''  multicast group management functions

29. Network Layer – part 329 Transition From IPv4 To IPv6  Not all routers can be upgraded simultaneously  no flag days  How will the network operate with mixed IPv4 and IPv6 routers?  Two proposed approaches:  Dual Stack: some routers with dual stack (v6, v4) can translate between formats  Tunneling: IPv6 carried as payload in IPv4 datagram among IPv4 routers

30. Network Layer – part 330 Dual Stack Approach

31. Network Layer – part 331 Tunneling IPv6 inside IPv4 where needed