Unit 9. Author: W.Buchanan. Routing Protocols (1)

Transcription

1 Unit 9 uthor: W.Buchanan. Routing Protocols (1)

2 Routing Protocols B H F B D E G C uthor: W.Buchanan. Routing Protocols (2)

3 Mobile s (Unit 10) Router Programming (Unit 8) Routing Protocols (Unit 9) IP ddressing/subnets (Unit 6) TCP/Socket Programming (Unit 7) Switches/VLNs (Unit 5) Net Elements (Unit 2) Net Types (Unit 3) Net Design (Unit 4) Introduction (Unit 1) uthor: W.Buchanan. Routing Protocols (3)

4 lternative Routes 1 Net1 Net2 2 Net5 Net4 4 Net6 6 Net8 BB 3 Net3 5 Net BB BB BB BB uthor: W.Buchanan. Routing Protocols (4)

5 Best route? Routing based on hops: Route (1,3,5,6) = 4 hops [BEST] Route (1,3,5,2,4,6) = 6 hops Routing based on delay (latency): Route(2,4,6) = = 2.75 Route(2,5,6) = = 2.4 [BEST] Routing based on error probability: P e (2 5)=0.01 P e (5 6)=0.15 P e (2 4)=0.05 P e (4 6)=0.1 P noerror (2,5,6) =(1 0.01) (1 0.15) = P noerror (2,4,6) =(1 0.05) (1 0.1) = [BEST] uthor: W.Buchanan. Routing Protocols (5)

6 Best Route Parameters? Bandwidth. The data capacity of a link, which is typically defined in bps. Delay. The amount of time that is required to send a packet from the source to a destination. Load. measure of the amount of activity on a route. Reliability. Relates to the error rate of the link. Hop count. Defined by the number of routers that it takes between the current router and the destination. Ticks. Defines the delay of a link by a number of ticks of a clock. Cost. n arbitrary value which defines the cost of a link, such as financial expense, bandwidth, and so on. uthor: W.Buchanan. Routing Protocols (6)

7 Best Route based on hops B H F B C D E G BH BCDFGH BEH 3 hops 7 hops 4 hops Unfortunately hop count may not be the best way to find the best route, as BH might involve a slow (or unreliable) link, such as BH, where BEH could be faster (or more reliable). uthor: W.Buchanan. Routing Protocols (7)

8 Dijkstra s lgorithm nalyse a part of a route that goes from one routing node to another, and reject the worst routes, so that it reduces the complexity of the problem 80 miles Stirling This route would be dismissed Falkirk Glasgow 64 miles Edinburgh uthor: W.Buchanan. Routing Protocols (8)

9 Dijkstra s lgorithm Inverness Oban 120 Stirling 20 Dunfermline Falkirk Glasgow Edinburgh Glasgow to Falkirk Glasgow to Oban Oban to Inverness Oban to Stirling Inverness to Stirling Stirling to Falkirk Stirling to Edinburgh Stirling to Dunfermline Falkirk to Edinburgh Dunfermline to Edinburgh 50 miles 100 miles 90 miles 120 miles 100 miles 5 miles 60 miles 20 miles 70 miles 15 miles uthor: W.Buchanan. Routing Protocols (9)

10 Dijkstra s lgorithm Inverness Oban Stirling 20 Dunfermline Falkirk 70 Glasgow Edinburgh uthor: W.Buchanan. Routing Protocols (10)

11 Dijkstra s lgorithm Inverness Oban Stirling 20 Dunfermline Falkirk 70 Glasgow Edinburgh uthor: W.Buchanan. Routing Protocols (11)

12 Inverness Oban Stirling Falkirk Glasgow Dunfermline 15 Edinburgh Edin, Falkirk, Glasgow: Edin, Dunf, Stir, Falk, Glasgow: 120 miles 90 miles uthor: W.Buchanan. Routing Protocols (12)

13 Layer 3 protocols Routing protocols. routing protocol provides a mechanism for routers to share routing information. These protocols allow routers to pass information between themselves, and update their routing tables. Examples of routing protocols are Routing Information Protocol (RIP), Interior Gateway Routing Protocol (IGRP), Enhanced Interior Gateway Routing Protocol (EIGRP), and Open Shortest Path First (OSPF). Routed protocols. These protocols are any network layer protocol that allows for the addressing of a host and a destination on a network, such as IP and IPX. Routers are responsible for passing a data packet onto the next router in, if possible, an optimal way, based on the destination network address. The definition of an optimal way depends on many things, especially its reachability. With IP, routers on the path between a source and a destination, examine the network part of the IP address to achieve their routing. Only the last router, which is connected to the destination node network, examines the host part of the IP address. uthor: W.Buchanan. Routing Protocols (13)

14 Types of Routing Dynamic routing. In dynamic routing, the routers monitor the network, and can change their routing tables based on the current network conditions. The network thus adapts to changing conditions. Unfortunately, this method tends to reveal everything known about an internetwork to the rest of the network. This may be inappropriate for security reasons. Static routing. In static routing, a system administrator sets up a manual route when there is only one route to get to a network (a stub network). This type of configuring reduces the overhead of dynamic routing. Static routing also allows the internetwork administrator to specify the information that is advertised about restricted parts of a network. Default routing. These are manually defined by the system administrator and define the path that is taken if there is not a known route for the destination. uthor: W.Buchanan. Routing Protocols (14)

15 Routing protocol types Distance-vector. Distance-vector routing uses a distance-vector algorithm (such as the Bellman-Ford routing algorithm), which uses a direction (vector) and distance to any link in the internetwork to determine the best route. Each router periodically sends information to each of its neighbors on the cost that it takes to get to a distance node. Typically, this cost relates to the hop count (as with RIP). The main problem with distance-vector is that updates to the network are step-bystep, and it has high bandwidth requirements as each router sends its complete routing table to all of its neighbors at regular intervals. Link-state. Link-state involves each router building up the complete topology of the entire internetwork (or at least of the partition on which the router is situated), thus each router contains the same information. With this method, routers only send information to all of the other routers when there is a change in the topology of the network. Linkstate is also known as shortest path first. Typical link-state protocols are OSPF, BGP and EGP. With OSPF, each router builds a hierarchical topology of the internetwork, with itself at the top of the tree. The main problem with link-state is that routers require much more processing power to update the database, and more memory as routers require to build a database with details of all the routers on the network. uthor: W.Buchanan. Routing Protocols (15)

16 Type of Update? Broadcast. In broadcast, routers transmit their information to other routers at regular intervals. typical broadcast routing protocol is RIP, in which routers send their complete routing table once every few minutes, to all of their neighbors. This technique tends to be wasteful in bandwidth, as changes in the route do not vary much over short amounts of time. Event-driven. In event-driven routing protocols, routing information is only sent when there is a change in the topology or state of the network. This technique tends to be more efficient than broadcast, as it does not use up as much bandwidth. uthor: W.Buchanan. Routing Protocols (16)

17 Routing protocol types Hybrid (IS-IS) Layer Layer 3 protocols Types Types Link-state Distance-vector Routed (IP, IPX, NetBEUI) Routing (RIP, OSPF) Session Session Transport Transport Data Data link link Physical Physical HTTP HTTP TCP TCP IP IP RIP RIP Ethernet/ Ethernet/ FDDI FDDI Routing Distance metrics Updates Hop count Delay Tick Bandwidth Cost Reliability Each router Each router transmits routing periodically sends information to information to all other routers each of its neighbors only when there (RIP). are changes (OSPF/BGP/EGP) Problems: Bandwidth Problems: Step-by-step updates Initial flooding Processing/memory Event driven v. broadcast Static.v. dynamic uthor: W.Buchanan. Routing Protocols (17)

18 Example routing Dest Hops Next Dest Hops Next 1 B 2 C 1 x z z 0 B 1 C 2 y y W 2 X 1 3 Dest Hops 2 B 1 C 0 Next w y C Z 4 Y Dest Hops 1 B 0 C 1 Next x B z C B uthor: W.Buchanan. Routing Protocols (18)

19 Routing loops Timing of events B B E. E. reachable reachable C D E E B. B. I I can can reach reach in in 3 3 hops hops W Z X Y.. unreachable unreachable D. D. reachable reachable Router Z thinks it can reach in 4 hops, as Router W says it can reach it in 3 hops, this overrules the information from Router Y which says it cannot reach.. unreachable unreachable C. C. Reachable Reachable via via Router Router W V unreachable uthor: W.Buchanan. Routing Protocols (19)

20 Overcoming Distance Vector Problems Setting infinity values. The count-to-infinity will eventually resolve itself when the routers have counted to infinity (as infinity will be constrained with the maximum definable value), but while the network is counting to this value, the routing information will be incorrect. To reduce the time that it takes to get to this maximum, a maximum value is normally defined. In RIP this value is set at 16 hops for hop-count distance-vectors, thus the maximum number of hops that can occur is 15. This leads to a problem in that a destination which has a distance of more than 15 hops is unreachable, as a value of 16 or more defines that the network is unreachable. Split horizon. This method tries to overcome routing loops. With this routers do not update their routing table with information on a destination if they know that the network is already connected to the router (that is, the router knows more about the state of the network than any other router, as it connects to it). Thus in Figure X, Router Z and Router X will not send routing information on B to Router Y, as they know that B is connected to Router Y. uthor: W.Buchanan. Routing Protocols (20)

21 Overcoming Distance Vector Problems Hold-Down Timers. This method overcomes the count-to-infinity problem. With a hold-time time, a router starts a hold-time timer when it receives an update from a neighbor indicating that a previously accessible network is now inaccessible. It also marks the route as inaccessible. There are then three possible situations: o If, at any time before the hold-down timer expires, an update is sent from the same neighbor which alerted the initial problem saying that it is now accessible, the router marks the network as accessible and removes the hold-down timer. o If an update arrives from a different neighboring router with a better metric than the original metric, the router marks the network as accessible and removes the hold-down timer. o If, at any time before the hold-down timer expires, an update is sent from a different neighbor which alerted the initial problem saying that it is accessible, but has a poorer metric than the previously recorded metric, the update is ignored. Obviously after the timer has expired the network will still be prone to looping routes, but the timer allows for a longer time for the network to settle down and recover the correct information. uthor: W.Buchanan. Routing Protocols (21)

22 Link-state overview 1 becomes unreachable for a short time LSP (Link state packets) Topological database (for SPF) Methods Problem Link-state LSP: LSP: LSP: LSP: Reachable Reachable Unreachable Unreachable W Z X LSP: LSP: Unreachable Unreachable LSP Operation 4 Y OSPF OSPF (RFC1583) (RFC1583) Ver. Ver. Type Type Message Message Len. Len. Router Router ID ID Concerns rea rea ID ID Checksum Checksum uth. uth. Type Type uthentication uthentication Each router Processing Memory builds up a tree Increased processing Increased topology of the power required to amount of subnetworks and find build trees storage memory shortest path for tree unreachable arrives after network reachable change in topology causes updates to all other routers uthor: W.Buchanan. Routing Protocols (22)

23 OSPF overview OSPF header OSPF OSPF (RFC1583) (RFC1583) Ver. Ver. Type Type Message Message Len. Len. Router Router ID ID (unique (unique in in S) S) rea rea ID ID (similar (similar to to subnetting) subnetting) Checksum Checksum uth. uth. Type Type uthentication uthentication dditional Information (depends on packet type) 32 bits Gateways OSPF is an IGP (Interior Gateway Protocol) which distributes routing information between routers in a single autonomous system. ll routers have the same database. Hello [1]. Used to establish and maintain a connection. Routers agree HelloIntervaland RouterDeadInterval. HelloInterval. Number of seconds between Hello packets. The smaller the value, the fastest the detection of topological changes. X.25 uses 30 sec, LNs uses 10 sec. RouterDeadInterval. Number of seconds before a router assumes that a route is down. It should be a multiple of HelloInterval (such as four times). Database Description [2]. Used to send database between routers. Link-state Request [3]. Request parts of a neighbor s database, which may be more up-to-date. Link-state Update [4]. Used to flood link state advertisements. Link-state cknowledgement[5]. Used to acknowledge flooded advertisements. Separate domains utonomous utonomous System System utonomous utonomous utonomous utonomous System System System System Internet EGP used between S s uthor: W.Buchanan. Routing Protocols (23)

24 Tree-like topology v. Internet-like topology Single backbone Single backbone uthor: W.Buchanan. Routing Protocols (24)

25 Example of RIP programming and IGRP Router# config t Router(config)# router rip Router(config-router)# network Router(config-router)# network Router(config-router)# network Router(config)# exit Router# exit S number Router broadcast their routing tables into these networks Router# config t Router(config)# router igrp 111 Router(config-router)# network Router(config-router)# network Router(config-router)# network Router(config)# exit Router# exit Every router within the domain must be running the same routing protocol uthor: W.Buchanan. Routing Protocols (25)