The University of Sydney 1 Routing Architecture and Protocols NETS3303/3603 Week 6

The University of Sydney 2 Outline Intro concepts –Issues –Architecture Routing protocols: –Distance-vector –Link-state Autonomous System and Exterior Gateway Protocol –BGP Interior Gateway Protocol –RIP1 and RIP2 –OSPF

The University of Sydney 3 Review - Internet Routing IP implements datagram forwarding Both hosts and routers –Have an IP module –Forward datagrams IP forwarding is table-driven –Table known as routing table

The University of Sydney 4 Elements of Routing routing protocols that allow info to be gathered and distributed – routers communicate with these protocols routing algorithms – compute good routes based on gathered data (like Bellman-Ford and Dijkstra) routing table – database of routes

The University of Sydney 5 Its Classic problem!

The University of Sydney 6 How / When Are IP Routing Tables Built? Depends on size / complexity of internet Static routing –Fixes routes at boot time –Useful only for simplest cases Dynamic routing –Table initialized at boot time –Values inserted / updated by protocols that propagate route information –Necessary in large internets

The University of Sydney 7 Routing Tables Two sources of information –Initialization (e.g., from disk) –Update (e.g., from protocols) Hosts tend to freeze the routing table after initialization But, routers use protocols to learn new information and update their routing table dynamically

The University of Sydney 8 Original Arpanet Routing Architecture Small set of ‘‘core’’ routers with complete information about all destinations Other routers know local destinations and use the core as central router (default route) Disadvantages of original core: –Central bottleneck for all traffic –No shortcut routes possible –Does not scale

The University of Sydney 9 General Idea – Better! Have a set of core routers know routes to all locations Devise a mechanism that allows other routers to contact the core to learn routes (spread necessary routing information automatically) Continually update routing information

The University of Sydney 10 Automatic Route Propagation Two basic algorithms used by routing update protocols –Distance-vector –Link-state Many variations in implementation details

The University of Sydney 11 Distance-Vector Algorithm Initialize routing table with one entry for each directly connected network Periodically run a distance-vector update to exchange information with routers that are reachable over directly connected networks Each router sends list of its routes to another

The University of Sydney 12 DV algorithm examples: RIP, BGP Its algorithm elements: –send: every N seconds out all connected interfaces broadcast 2-tuples : (to network X, hop count Y)... –recv: if new tuple, add to routing table; if better tuple, change existing; if “dead” tuple, remove –timeout: if no refresh, timeout entry in N * Y seconds

The University of Sydney 13 Example Of DV Update Router K received an update from router J (a) is existing routing table at K (b) incoming update (marked items cause change)

The University of Sydney 14 slow convergence/count to infinity DV’s big problem! changes can be sent when they occur, but must recompute a bit so convergence takes time (made worse by possible loops) count to infinity problem can occur too - routing loop until hopcount reaches impossible value

The University of Sydney 15 Count to infinity C crashes, B knows C crashed but hasn’t told A, but unfortunately A talks to B first B is told by A: –I can get to C in two hops (and note it doesn’t mention to B that the path is thru B) B says AHA!, that means I can get to C in three hops and reports that to A A says AHA!, it’s now four hops to B and tells B etc...

The University of Sydney 16 split-horizon and poison reverse fixup Split-horizon: –A does not tell B that it can reach C (white lie) –Because its through B Poison reverse: –When B loses connection to C, its distance to C is changed to infinity –An immediate update is triggered, without wait for regular update when link goes away, B will know that there is no path to C, and tell A Still doesn’t work in all cases…

The University of Sydney 17 Link-State Algorithm Alternative to distance-vector Distributed computation –Broadcast information –Allow each router to compute shortest paths Avoids problem where one router can damage the entire internet by passing incorrect information Also called Shortest Path First (SPF)

The University of Sydney 18 Link-State Update Participating routers learn internet topology Think of routers as nodes in a graph, and networks connecting them as edges or links Pairs of directly-connected routers periodically –Test link between them –Propagate (broadcast) status of link All routers –Receive link status messages –Recompute routes from their local copy of information A E D CB F

The University of Sydney Determine link-state

The University of Sydney Send LS-update

The University of Sydney Compute shortest path

The University of Sydney 22 link-state: pros/cons pros –converges faster, no count to infinity problem + router can forward LSP immediately –more functionality; e.g., each router has map of net, can make network debugging easier cons –more compute than DV (does this matter?)

The University of Sydney 23 ROUTING: EXTERIOR GATEWAY PROTOCOLS AND AUTONOMOUS SYSTEMS (BGP)

The University of Sydney 24 General Principle for Internet Routing Although it is desirable for routers to exchange routing information, it is impractical for all routers in an arbitrarily large internet to participate in a single routing update protocol Consequence: routers must be divided into groups

The University of Sydney 25 A Practical Limit On Group Size Up to a dozen routers to participate in a single routing area across a WAN approximately five times as many can safely participate across a set of LANs

The University of Sydney 26 Router Outside A Group Does not participate directly in group’s routing information propagation algorithm Problems: –Will not choose optimal routes if it uses a member of the group for general delivery –May not know all networks from other groups

The University of Sydney 27 The Extra Hop Problem Non-participating router picks one participating router to use (e.g., R 2 ) Non-participating router routes all packets to R 2 across backbone Router R 2 routes some packets back across backbone to R 1 So, a mechanism is needed that allows nonparticipating routers to learn routes from participating routers –so they can choose optimal routes.

The University of Sydney 28 The Hidden Networks Problem Group must learn routes from nonparticipating routers Example: owner of networks 1 and 3 must tell group that there is a route to network 4

The University of Sydney 29 Autonomous System Concept (AS) Group of networks under one administrative authority Free to choose internal routing update mechanism Connects to one or more other autonomous systems

The University of Sydney 30 EGPs: Exterior Gateway Protocols A protocol for communicating routes between two autonomous systems Solves two problems –Allows router outside a group to advertise networks hidden in another autonomous system –Allows router outside a group to learn destinations in the group

The University of Sydney 31 Border Gateway Protocol The most popular (virtually the only) EGP in use in the Internet Current version is BGP-4 Supports CIDR (mask accompanies each route) Each AS designates a border router to speak on its behalf Two border routers become BGP peers

The University of Sydney 32 Key Characteristics Of BGP Provides inter-autonomous system communication Propagates reachability information Follows next-hop paradigm Provides support for policies Sends path information Permits incremental updates Allows route aggregation Allows authentication

The University of Sydney 33 Additional BGP Facts Uses reliable transport (i.e., TCP) –Unusual: most routing update protocols use connectionless transport (e.g., UDP) Sends keepalive messages so other end knows connection is valid (even if no new routing information is needed)

The University of Sydney 34 Four BGP Message Types

The University of Sydney 35 BGP Message Header Each BGP message starts with this header Marker is used by peers to indicate message boundary => synchronisation

The University of Sydney 36 BGP Open Message Used to start a connection HOLD TIME specifies max time that can elapse between BGP messages

The University of Sydney 37 BGP Update Message Sender can advertise new routes or withdraw old routes Each route entry consists of address and mask Entry can be compressed to eliminate zero bytes

The University of Sydney 38 BGP Must Consider Receiver’s Perspective Two issues are considered: policies and optimal routes Advertise not just destinations but report reachability too

The University of Sydney 39 Path Metric Interpretation Each AS use own IGP may be with different metric (hop count, delay, policy-based values) So, an exterior gateway protocol does not communicate or interpret metrics, even if metrics are available! BGP only propagates reachability information –a receiver can implement policy constraints –BUT cannot choose a least cost route.

The University of Sydney 40 ROUTING: INSIDE AN AUTONOMOUS SYSTEM (RIP, OSPF) Static routes –Initialized at startup –Never change –Typical for host –Sometimes used for router Dynamic routes –Initialized at startup –Updated by route propagation protocols –Typical for router –Sometimes used in host

The University of Sydney 41 Exchanging Routing Information Within An Autonomous System Mechanisms called interior gateway protocols, IGPs Choice of IGP is made by autonomous system

The University of Sydney 42 Example Of Two Autonomous Systems And the Routing Protocols Used RIPOSPF

The University of Sydney 43 Routing Information Protocol (RIP) Implemented by UNIX program routed Uses hop count metric Distance-vector protocol Relies on broadcast Assumes low-delay LAN Uses split horizon and poison reverse techniques to solve inconsistencies Current standard is RIP2

The University of Sydney 44 Two Forms Of RIP Active –Form used by routers –Broadcasts routing updates periodically –Uses incoming messages to update routes Passive –Form used by hosts –Uses incoming messages to update routes –Does not send updates

The University of Sydney 45 RIP Operation Each router sends update every 30 seconds Update contains pairs of (destination address, distance) Distance of 16 is infinity (i.e., no route) –This limits span of its internet to 16 hops –Any 2 nodes have at most 15 routers between

The University of Sydney 46 Illustration Of Hosts Using Passive RIP Host listens for RIP broadcast and uses data to update table Eliminates ICMP redirects Host routing table initialized to: DestinationRoute default Direct

The University of Sydney 47 Changes To RIP In Version 2 RIP1 does not include subnet mask –Only suitable for classful or fixed-len subnets Update includes subnet mask Authentication supported Explicit next-hop information Messages can be multicast (optional) –IP multicast address is

The University of Sydney 48 RIP2 Update Format

The University of Sydney 49 Open Shortest Path First (OSPF) Developed by IETF in response to vendors’ proprietary protocols Uses link-state algorithm More powerful than most predecessors Permits hierarchical topology More complex to install and manage

The University of Sydney 50 OSPF Features Type of service routing – the first to offer Load balancing across multiple paths Networks partitioned into subsets called areas –Designated router per area Message authentication Virtual network topology abstracts away details Can import external routing information

The University of Sydney 51 OSPF Message Header Each message starts with same header OSPF Message Types –1 Hello (used to test reachability) –2 Database description (topology) –3 Link status request –4 Link status update –5 Link status acknowledgement

The University of Sydney 52 OSPF HELLO Message Format Used to establish and test reachability

The University of Sydney 53 OSPF Database Description Message Format Initialises network topology database One serve as a Master; other slave Can be large => separate into different msgs –I is 1 for first message, M is 1 for more messages Router Network Summary of nets AS boundary

The University of Sydney 54 Summary Internet is too large for all routers to participate in one routing update protocol Group of networks and routers under one administrative authority is called Autonomous System (AS) EGP is used to communicate routing information between two autonomous systems Each AS chooses its own interior routing update protocol Popular IGPs include –RIP (distance vector algorithm) –OSPF (link-state algorithm)