ROUTING PROTOCOL BASICS

Transcription

1 ROUTING PROTOCOL BASICS SCOTT KIRBY SHANKAR VEMULAPALLI 1 Introduction Provide a general view of four routing protocols which include RIP, EIGRP, OSPF and IS-IS. The goal is to show the common characteristics between the protocols and give the student enough knowledge to compare and contrast them 2 1

2 Agenda What does a Router do? Distance Vector Routing Protocol Advanced Distance Vector Routing Protocol Link State 3 Overview How do you direct traffic in an efficient manner across the network while preventing loops? 4 2

3 WHAT DOES A ROUTER DO? Presentation_ID 2003, 2004 Cisco Systems, Inc. All rights reserved. 5 What Does a Router Do? What s going on in here? Serial 1/0 6 3

4 What Does a Router Do? 7 Strips Datalink Header Don t need this anymore! The data link header already did its job since the packet has arrived Data Dst IP Addr Src IP Addr CS/Prot TTL Misc TOS IHL/Ver DataLink Hdr Ethernet 8 4

5 Decrement Time-to-Live by One Mitigation necessary as insurance against routing loops The TTL value is decremented by 1 in each router that it traverses. Packet is dropped if TTL reaches zero Data Dst IP Addr Src IP Addr CS/Prot TTL Misc TOS IHL/Ver Looks for Destination IP Subnet Data Dst IP Addr Src IP Addr CS/Prot TTL Misc TOS IHL/Ver 10 5

6 Matches the Subnet to Forwarding Information Router1#sh ip route Codes: C connected, D - EIGRP D C D D D /8 is variably subnetted, 3 subnets, 2 masks /30 [90/ ] via , 00:01:32, Serial1/ /30 is directly connected, Serial1/ /27 [90/ ] via , 00:04:38, Serial1/ /16 [90/ ] via , 00:02:45, Serial1/ /8 [90/ ] via , 00:01:29, Serial1/0 Data Dst IP Addr Src IP Addr CS/Prot TTL Misc TOS IHL/Ver 11 Must Be Loop Free The routing protocol should give me a loop free route Data Dst IP Addr Src IP Addr CS/Prot TTL Misc TOS IHL/Ver 12 6

7 Datalink Header Rewrite with Next Hop Info A new datalink header is reapplied after consulting the forwarding information base D /30 [90/ ] via , 00:01:32, Serial1/0 Data Dst IP Addr Src IP Addr CS/Prot TTL Misc TOS IHL/Ver DataLink Hdr PPP 13 How Does the Router Get the Information? Static Routing or Routing Protocols The goal is to find a loop free path to the destination We will discuss two types of Routing Protocol Algorithms: Distance Vector Link State 14 7

8 Static Routing Static Routes are entered on each router by the network administrator The Command syntax is: ip route target network (tn) mask next-hop-address The Target Network is an IP Prefix and the mask represents how many bits are representing the network portion of the address The Next-Hop-Address identifies the interface to use and the address of the next hop downstream to the destination 15 Installing a Route in the Forwarding Table After a router has been presented with routes to a destination from Static Routing or Dynamic Routing, the route with the lowest Administrative Distance is selected and used as the best path to the destination Administrative Distance gives the router an ability to prefer routes from particular routing protocols (or Static Routing) over other protocols Router1#sh ip route Codes: C connected, D - EIGRP D /8 is variably subnetted, 3 subnets, 2 masks /30 [90/ ] via , 00:01:32, Serial1/0 16 8

9 Illustrate Routing Table with Static Routes Show IP route Enter static routes Show IP route again 17 DISTANCE VECTOR ROUTING PROTOCOLS 18 9

10 Bellman-Ford The original algorithms for distance vector routing were based on the works of R.E. Bellman, L.R. Ford, and D.R. Fulkerson. Therefore often referred to as Bellman-Ford or Ford-Fulkerson The most common implementation of the Bellman- Ford algorithm is in the Routing Information Protocol (RIP) protocol 19 Distance Vector Algorithms Routing Information Protocol (RIP) and Interior Gateway Routing Protocol (IGRP) RIP and IGRP using the Bellman-Ford Algorithm take a list of available routes with associated simple metrics and selects the one with the lowest metric while discarding the others Enhanced Interior Gateway Routing Protocol EIGRP using the Diffusing Update Algorithm takes a list of neighbor s routes with associated complex metrics and stores them in order to pick a route that is guaranteed to be loop free and provide the most efficient route 20 10

11 Bellman-Ford Distance Vector Routing Protocols Xerox Networking System s XNS Routing Information Protocol (RIP) Novell s IPX Routing Information Protocol (RIP) Cisco s Internet Gateway Routing Protocol (IGRP) Appletalk s Routing Table Maintenance Protocol (RTMP) IP Routing Information Protocol (RIP) 21 IP RIP Benefits First used for Xerox PUP Protocol IP version of RIP was based on a program in Berkeley Unix called routed and was considered the de facto standard for IP routing RIP was formally documented in RFC 1058 Designed for small to medium sized networks Can also be easily optimized for use in networks with many neighbors such as Frame Relay or ATM Very easy configuration 22 11

12 Information Needed for Each Route With the following information and updates from neighbors, RIP can calculate a path through the network IP Network Prefix of the target network IP Address of first router (next hop) Interface used to reach the next hop to the target Metric which indicates the distance to the target network 23 Router Router Initialization Routers start with a database (Routing Information Base) of directly connected routes The database might contain static routes entered by the network administrator The database is then updated according to information received in update messages from neighboring routers The messages from the neighbors reflect their Forwarding Information Base which is the best path to the most destinations from their perspective 24 12

13 RIP Routing Updates All routers send updates (every 30 seconds) to each other showing routes that they can access This illustration will only focus on the a single network (Target Network) for simplicity Routers add a cost of 1 to the route they have learned when re-advertising it 25 RIP Routing Update Contents 4 Bytes Command Version Routing Domain Address Family ID Route Tag IP Address Subnet Mask Next Hop Metric 2(Update) 2 (ver) 0 (Default) 2 (IP) 0 (Default) (TN-Target Network) (Specifies the originator) 0 UDP Packet to and from Port 520 May appear up to 25 times in a single 512 byte packet RIP v2 Additions 26 13

14 RIP Routing Update Contents for Route to TN 2(Update) 2 (ver) 0 (Default) 2 (IP) 0 (Default) (TN-Target Network) (Specifies the originator) 0 27 Next Hop Address for Route to TN 2(Update) 2 (ver) 0 (Default) 2 (IP) 0 (Default) (TN-Target Network) (Specifies the originator) 0 If the next hop address shows that indicates to the neighboring router that the next hop to use should be the source address of the IP packet If this were a multi-access network, the next hop address would prevent packets from being routed through extra hops 28 14

15 Metric for Route to TN 2(Update) 2 (ver) 0 (Default) 2 (IP) 0 (Default) (TN-Target Network) (Specifies the originator) 0 The metric indicates the number of routers that must be traversed in order to reach the Target Network. Since D is advertising the route and is directly attached, the metric is advertised as 0. B will add 1 to this metric when re-advertising it 29 Metric The route with the best metric is assumed to be loop free. Other routes with higher metrics are assumed to be possible loops With RIP this is generally a hop count. Other protocols such as EIGRP contains much more information in the metric such as minimum bandwidth and total delay RIP has a maximum metric of

16 Important Update Rule A common rule among Distance Vector protocols is that a router must only advertise routes that it would use to reach a destination itself 31 Timers Every 30 seconds a complete update is sent to all neighboring routers Timeout (180 seconds without an update for a route) is used to invalidate routes while retaining the route for a short time to notify neighbors that the route has been dropped by marking the route to a metric of infinity Garbage Collection Timer (120 seconds starting after Timeout) to remove routes from table (Route continues to be advertised with a metric of infinity until the Garbage Collection time expires). Also referred to in IGRP as the Flush Timer 32 16

17 RIP Link Metric Adjustments Metrics can be changed by the administrator on certain links for routing preference Here the link between C and D is set to a metric of 10 because it is a slower interface than the links between the other routers This action does reduce the overall diameter of the network from 15 to 6 hops Hops are the number of routers being traversed 33 Route Advertisements from D The illustration will only show the metric for TN being advertised in the update D advertises all routes stored in its routing table (although we are only showing TN) Router C and B install the newly advertised route (plus the metric of the incoming interface used) in their routing table 34 17

18 Route Computation in B B receives TN with the metrics of used by D and then adds the metrics from the incoming 10BaseT port which is 1 All Links are 10BaseT except C to B which is 512K Total Metric = (Incoming Metric in the Update from D for TN) 0 + (Metric of Ethernet0 port between B and D) 1 = 1 RIP would store TN with a metric of 1 into the routing table 35 Route Advertisements from B B then advertises TN with a metric of 1 to A, C and back to D (which ignores it) C installs the better route and adds the incoming link metric to it (2 is better than 10) and discards the previous route A installs the newly discovered route (with the appropriate metric of 1 + incoming link metric) 36 18

19 Route Computation in B B receives TN with the metrics of used by D and then adds the metrics from the incoming Ethernet port which is 1 All Links are 10BaseT except C to B which is 512K Total Metric = (Incoming Metric in the Update from B for TN) 1 + (Metric of Ethernet0 port between B and D) 1 = 2 RIP would store TN with a metric of 2 into the routing table 37 Route Advertisements from A A advertises TN with metric of 2 to B and C Routers B and C ignore it since they have a better metric for TN 38 19

20 Route Advertisements from C C advertises TN with metric of 2 to A, B and D Routers A, B and D ignore it since they have a better metric installed 39 Converged Path Converged Any packets destined for the TN network will follow the paths illustrated Routes to other destinations may or may not follow this path 40 20

21 B to D Link Failure If the link between B and D fails, we will see how long it could take for RIP to converge New path should be between C and D Note: if the B notices the link failure, the routes will be flushed immediately. We are assuming that B did not notice the link failure X 41 RIP Convergence Process C still advertises TN every 30 seconds with metric of 2 to A, B and D A will not use it until the existing route times out (180 seconds) If B noticed a hardware failure then the route would have been timed out immediately and would install the route from C with a metric of 3 X 42 21

22 RIP Convergence Process B now advertises TN with a metric of 3 to A and C A and C see that the metric through B has changed so they install a new metric of 4 X 43 RIP Convergence Process A advertises TN with a metric of 4 to B and C B and C will ignore it since the metric is higher than the route installed X 44 22

23 RIP Convergence Process C also advertises TN with a metric of 4 to A, B and D B sees that the metric has changed via C and updates its route A ignores it since the metric is higher than the route installed X 45 RIP Convergence Process B advertises TN with a metric of 5 to A and C Updates continue until the metric from B reaches 11 X 46 23

24 RIP Convergence Process C learns that B increased its metric to TN and ignores it since D is advertising a lower metric C uses the route advertised from D which is 10 (after adding the link metric) A updates its metric to 12 X 47 RIP Convergence Process C advertises TN with a metric of 10 and to A, B and D A and B update their routing table with the new metric X 48 24

25 RIP Converged around Failure Finally Converged During this convergence, any packets destined for TN would loop between C and B until their TTL reached 1 X 49 Counting to Infinity If the link between C and D did not exist, the counting would continue to infinity Packets destined to TN will loop between A, B and C until the convergence occurs That is why RIP has a 15 hop count maximum to reduce how long a loop can exist X 50 25

26 Additional Convergence Mechanisms The following mechanisms will help out: Simple split horizon Split Horizon with Poisoned Reverse Triggered Updates X 51 RIP Convergence with Split Horizon, Poison Reverse and Triggered Updates Split horizon omits routes learned from one neighbor in updates sent to that neighbor Poison Reverse poisons (sets metric to 15) the route in updates to neighbors instead of omitting that route Triggered Updates cause router to send updates almost immediately when the metric on a route changes (Also called flash updates) 52 26

27 RIP Convergence with Split Horizon, Poison Reverse and Triggered Updates B notices the link failure and sends a triggered update (TN update with metric of 15) to A and C (slightly staggered in this example) A gets the update first and removes its route to TN Meanwhile C has reached its update interval before receiving B s Update so C advertises TN with a metric of 3 to A and D C then receives the update from B at a metric of 15 and updates the route, but it s too late X 53 RIP Convergence with Split Horizon, Poison Reverse and Triggered Updates C does send a triggered / flash updates to C with a metric of 16, but A has managed to already send an update to B B will install that route X 54 27

28 RIP Convergence with Split Horizon, Poison Reverse and Triggered Updates After getting the flash update from C, A sends a flash update to B with a metric of 15 Since the flash updates are not waiting on a timer and normal updates are, B should get the infinity metric before sending an update which will cause the route to start the deletion process (garbage collection timer) X 55 RIP Convergence with Split Horizon, Poison Reverse and Triggered Updates C learns a new route from D and will install it since it has a better metric than the infinity metric through B X 56 28

29 RIP Convergence with Split Horizon, Poison Reverse and Triggered Updates C advertises TN with a metric of 10 to A and B A and B update their routing table with the new route X 57 RIP Convergence with Split Horizon, Poison Reverse and Triggered Updates Converged much faster with flash updates, split horizon and poison reverse Still a small chance of counting to infinity, but unlikely Still slow convergence due to the garbage collection timer of 120 seconds. Old routes need to be removed before new routes are learned X 58 29

30 RIP Convergence Lost Packet PROBLEM What happens if the triggered update packet from A was lost? RIP does not have a guaranteed delivery mechanism B will have time to send another routing update to C which will start the counting to infinity X X 59 RIP Problem Areas Network Diameter limit of 15 hops Vulnerable to counting to infinity with packet drops Not a very intelligent route decision process Excessive use of bandwidth due to periodic broadcast of routing updates 60 30

31 INTERIOR GATEWAY ROUTING PROTOCOL: IGRP Presentation_ID 2003, 2004 Cisco Systems, Inc. All rights reserved. 61 IGRP Improvements over RIP All Features of RIP are included in IGRP Plus: Increased network diameter by introducing an extended Metric, a Hold Down Timer, and improved Poisoning Designed for Small to Medium Networks Improved route decisions by adding Bandwidth, Delay, Load and Reliability information to the metric Load Balancing over roughly equal metric routes 62 31

32 IGRP Hold Down The diagram refers to the dropped packet issue in RIP IGRP implements a hold down timer (RIP also does now) After a route has been marked unreachable, hold down causes a router to sit on the route to TN for enough time for multiple update messages to cycle around the network This insures that everyone hears of a link failure before accepting new routing information X X 63 IGRP Timers Every 90 seconds a complete update is broadcast to all neighboring routers Timeout/Invalid Time (270 seconds without an update for a route) is used to invalidate routes while retaining the route for a short time to notify neighbors that the route has been dropped by marking the route to a metric of infinity Hold Time 280 seconds (no new path is learned after a path has gone unreachable (or the metric has increased enough to cause Poisoning) Garbage Collection/Flush Timer (630 seconds starting after hold time) to remove routes from table (Route continues to be advertised with a metric of infinity until the Garbage Collection time expires) 64 32

33 Route Poisoning Improvements Using RIP, the poison reverse capability poisons (sets metric to 15) the route in updates to neighbors when the link used for that route has disappeared In IGRP, when an update shows the metric for an existing route to have increased sufficiently (composite metric increases more than a factor of ~1.1), it could be a loop. The route should be removed and put into hold down 65 IGRP Metric Improvement The Metric is a Composite Metric which is calculated from variables of Bandwidth (minimum path bandwidth) Delay (total path delay) Reliability (minimum path reliability) Composite Delay is calculated by running the variables through an equation 66 33

34 IGRP Metric Equation To calculate the Composite Metric IGRP uses this equation: K1 bandwidth + ( K2 bandwidth) ( ) ( ) K5 + K3 Delay 256 Load Reliability + K 4 Default values for K1 and K3 are 1 and K2, K4, and K5 are 0 Delay is a composite delay measured in microseconds and then divided by 10 and multiplied by 256 Bandwidth is the minimum bandwidth of the path and is measured as the inverse (by dividing the 10,000 by the actual bandwidth and then multiplying by 256 (10,000,000 / minimum bandwidth of path) Reliability the minimum reliability of the path and is given as a fraction of 255 representing the percentage of packets that have been successfully transmitted and received with 255 being the most reliable. Same as the variable in show interface Load is the maximum of the load on the path and measured as a fraction of 255 with 1 being the least load. Same as the variable in show interface 67 IGRP Routing Update Contents Version Opcode Edition # Autonomous System Number Number of Subnets in local network Number of networks in AS Number of networks outside AS Checksum of IGRP Header and Data IP Prefix Delay Bandwidth MTU Reliability Load Hop Count Common Header for all IGRP Packets Internal Route Update information duplicated for each route Note that the Composite metric is not advertised since it does not contain the necessary information to use for a decision based on Bandwidth and Delay 68 34

35 Route Advertisements from D D advertises network TN, with a vector of metrics containing the Bandwidth, Delay, Load and Reliability which represent the Ethernet segment of D C and B add the delay and compare the bandwidth, load, and reliability based on the metrics of the link that the update arrived on C and B then convert the vector of metrics into a composite metric and use that to compare routes All links have a delay of 1msecs except between C and D which is 5msecs 69 Route Computation in B B receives TN with the metrics of the Ethernet port on D and then adds/ compares the metrics from the incoming Ethernet port All links have a delay of 1msecs except between C and D which is 5msecs Incoming Metric for TN: Bandwidth 10000, Delay 1000, Load 1, Reliability 255 Assuming (K1=K3=1, all other Ks are 0) the calculation would be the Min Inverse Scaled Bandwidth of (10M) + Scaled Delay of (2 msecs) = Composite Metric of IGRP would store TN with a composite metric of into the routing table 70 35

36 Route Computation in C C receives TN with the metrics of the Ethernet port on D and then adds/ compares the metrics from the incoming 512K port All links have a delay of 1msecs except between C and D which is 5msecs Incoming Metric for TN: Bandwidth 10000, Delay 1000, Load 1, Reliability 255 Assuming (K1=K3=1, all other Ks are 0) the calculation would be the Min Inverse Scaled Bandwidth of (512K) + Scaled Delay of (6 msecs) = Composite Metric of IGRP would store TN with a composite metric of into the routing table 71 Route Advertisements from C C will advertise TN with a Bandwidth of 512K (since it is the minimum so far on the path) and a composite delay of 6Msecs which is the Ethernet (TN) (1 msec) plus the 512K link (5 msec) between C and D All links have a delay of 1msecs except between C and D which is 5msecs 72 36

37 Route Computation in B B receives an update from C with the metrics stored on C then adds the metrics of the B to C link and computes the composite metric and compares it to the installed route All links have a delay of 1msecs except between C and D which is 5msecs Incoming Metric for TN: Bandwidth 512K, Delay 6Msecs, Load 1, Reliability 255 Assuming (K1=K3=1, all other Ks are 0) the calculation would be the Min Inverse Scaled Bandwidth of (512K) + Scaled Composite Delay of (7000 usecs) = Composite Metric of IGRP would ignore this route since it is greater than the composite metric of Route Advertisements from B B will advertise send an update with it s metrics for TN to A and C All links have a delay of 1msecs except between C and D which is 5msecs 74 37

38 Route Computation in C C receives the route from B and adds the metrics for the C to B link and computes the composite metric to use for a route comparison All links have a delay of 1msecs except between C and D which is 5msecs Incoming Metric for TN: Bandwidth 10000, Delay 2msecs, Load 1, Reliability 255 Assuming (K1=K3=1, all other Ks are 0) the calculation would be the Min Inverse Scaled Bandwidth of (10M) + Scaled Composite Delay of (3 msecs) = Composite Metric of That s a better route, so IGRP will install that as the best path to reach TN 75 IGRP Computation with Load and Reliability Adding Load and Reliability to the equation by setting the K2, K4 and K5 coefficient to 1 and increasing the Bandwidth K1 and Delay K3 coefficients to 5 to reduce the likelihood of routing loops based on too many errors and too much load NOTE: All K Coefficients must be the same in the network (aka Autonomous System)! Be VERY careful about changing the K values to insure that you don t cause a routing loop 76 38

39 Computation with K1=K3=5 and K2=K4=K5=1 When D advertises TN, the advertisement will contain a vector of metrics containing the Bandwidth, Delay, Load and Reliability C and B add the delay and adjust the bandwidth, delay and reliability based on the metrics of the link where the update arrived C and B then convert the vector of metrics into a composite metric using the new K values and then install it if it is the lowest metric for TN All links have a delay of 1msecs except between C and D which is 5msecs 77 Computation with K1=K3=5 and K2=K4=K5=1 C receives an update from D with D s vector metrics and then adds the C to D link metrics and calculates the composite metric with the new K values in route computation All links have a delay of 1msecs except between C and D which is 5msecs Incoming Metric for TN: Bandwidth 10Mbps, Delay=1msec, Load=1, Reliability=255 Using the new K1 values the calculation would be: (K1 x ScaleBandwidth+(K2 x ScaleBandwidth) /(256-Load)+K3 x ScaleDelay) x IF(K5=0,1,(K5/(Reliability+K4))) = IGRP would install the route to TN with the next hop of D using a composite metric of

40 Computation with K1=K3=5 and K2=K4=K5=1 B receives an update from D with D s vector metrics and then adds the B-to-D link metrics and calculates the composite metric with the new values in route computation All links have a delay of 1msecs except between C and D which is 5msecs Incoming Metric for TN: Bandwidth 10Mbps, Delay=1msec, Load=1, Reliability=255 With new K values the calculation would be: (K1 x ScaleBandwidth+(K2 x ScaleBandwidth) /(256-Load)+K3 x ScaleDelay) x IF(K5=0,1,(K5/(Reliability+K4))) = 6003 IGRP would place the route to TN with next hop of D and a composite metric of Computation with K1=K3=5 and K2=K4=K5=1 C receives TN and then adds/compares the metrics from the incoming Ethernet port All links have a delay of 1msecs except between C and D which is 5msecs Incoming Metric for TN: Bandwidth=10Mbps, Delay=2msecs, Load=1, Reliability=255 With the new K values the calculation would be: (K1 x ScaleBandwidth+(K2 x ScaleBandwidth) /(256-Load)+K3 x ScaleDelay) x IF(K5=0,1,(K5/(Reliability+K4))) = 6503 Since 6503 is lower than , IGRP will replace the route to TN with these new metric of

41 Reliability and Load Changes C receives TN and then adds/compares the metrics from the incoming Ethernet port All links have a delay of 1msecs except between C and D which is 5msecs Incoming Metric for TN: Bandwidth=10Mbps, Delay=2msec, Load=1, Reliability=15 With the new K values the calculation would be: (K1 x ScaleBandwidth+(K2 x ScaleBandwidth) /(256-Load)+K3 x ScaleDelay) x IF(K5=0,1,(K5/(Reliability+K4))) = IGRP will update the metric to TN using the composite metric of Reliability and Load Changes C receives another update from D and replaces the existing route that is experiencing errors All links have a delay of 1msecs except between C and D which is 5msecs Incoming Metric for TN: Bandwidth 10Mbps, Delay=1msec, Load=1, Reliability=255 Using the new K1 values the calculation would be: (K1 x ScaleBandwidth+(K2 x ScaleBandwidth) /(256-Load)+K3 x ScaleDelay) x IF(K5=0,1,(K5/(Reliability+K4))) = IGRP would install the route to TN with the next hop of D using a composite metric of since it is smaller than the previous low reliability metric 82 41

42 IGRP Problem Areas Slow Convergence around failures due to the loop prevention timers Uses more bandwidth than necessary due to periodic routing table broadcasts Unable to carry a default network of IGRP requires you to tag a real network as a default network 83 ENHANCING IGRP USING DIFFUSING UPDATE ALGORITHM (DUAL) Presentation_ID 2003, 2004 Cisco Systems, Inc. All rights reserved

43 EIGRP Benefits Increases the size of the metric from IGRP Replaces Bellman-Ford Algorithm with Diffusing Update Algorithm (DUAL) for loop free route calculations Designed for Small to Large networks Multi-protocol support adding AppleTalk and IPX Reliable Transport Protocol with Multicast which protects against lost packets Neighbor Adjacencies for quicker convergence Faster Convergence by establishing Neighbor Relationships, maintaining a topology database and using DUAL Incremental Update to save bandwidth and increase efficiency 85 EIGRP Metrics The Metric is the same metric used in IGRP x 256 which provides bandwidth, delay, load and reliability intelligence although the metric has been increased from 24 bits to 32 bits 86 43

44 EIGRP Topology Database EIGRP stores all routes that have been learned by its neighbors as opposed to picking the best and discarding the rest The routes are stored in the topology database with metrics as reported from the neighbors without adding the local link metric to reach the neighbor 87 Neighbor Process: Review Neighbor Relationships are maintained to insure that the routes in the topology database are usable Eliminates the periodic flooding Neighbor Relationships are common for Link State Routing Protocols and only used in one Distance Vector Protocol (EIGRP) A Hello B 88 44

45 EIGRP 3-Way Handshake During normal adjacency formation, multicast hellos cause the EIGRP process to place new neighbors in the neighbor table Unicast packets are then used to exchange known routing information, and complete the neighbor relationship multicast hello unicast update + init unicast acknowledgement unicast topology table info A B 89 EIGRP Timers A complete update is initially broadcast to all neighboring routers Neighbor Hello Interval How often the neighbors send a Hello to each other. Configurable between 5 and 60 seconds Hold Time Neighbor Timeout - 3 times the Hello Interval Multicast Flow Control Timer the amount of time the system will wait for all neighbors on a multicast interface to acknowledge the previous packet before attempting to unicast the same packet Active Timer control the time the router waits (after sending a query) before declaring the route to be in the stuck in active (SIA) state. The default is 10 minutes after the rewrite and 3 minutes before the rewrite 90 45

46 Checking Neighbor Status RTRA# show ip eigrp neighbors IP-EIGRP neighbors for process 1 H Address Interface Hold Uptime SRTT RTO Q Seq (sec) (ms) Cnt Num Et0 12 6d16h Et1 13 2w2d Et1 10 2w2d Hold How long to wait for an EIGRP packet before declaring this neighbor dead Uptime How long since last time this neighbor was discovered 91 Holding Time Expired Holding time expires when an EIGRP packet is not seen for the duration of the hold time Usually caused by missing multicast hello packets Typically caused by congestion or physical errors A Hello Neighbor (Ethernet0) is down: holding time expired B 92 46

47 DUAL Each of A s neighbors is reporting reachability to E: B with a cost of 10 A 15 D C with a cost of 10 D with a cost of These three costs are called the reported distance (RD) ; the distance each neighbor is reporting to a given destination B 10 C E 93 DUAL At A, the total cost to reach E is: 20 through B A 15 D 25 through C 45 through D The best of these three paths is the path through B, with a cost of 20 B 10 C This is the feasible distance (FD) E 94 47

48 DUAL A uses these two pieces of information to determine which paths are loop free The best path (Feasible Distance) is used as a benchmark; all paths with Reported Distances lower than the Feasible Distance cannot contain loops Although the algorithm may mark some loop free paths as loops, it is guaranteed never to mark a looped path as loop free B A C E D 95 DUAL At A: A The path through B is the best path (Feasible Distance), at D C can reach E with a cost of 10; 10 (Reported Distance) is less than 20 (Feasible Distance), so this path is loop free D can reach E with a cost of 20; 30 (Reported Distance) is not less than 20 (Feasible Distance), so EIGRP assumes this path is a loop B C E 96 48

49 DUAL If the best path fails, through B (the successor), EIGRP will examine the available paths to E A 15 D Finding a path which was previously declared loop free (a feasible successor), it begins using it immediately C now becomes the successor (best path) B C E 97 DUAL A now examines its topology information based on the new successor metric The reported distance through the remaining neighbor, D, is 30; 30 (Reported Distance) is still more than 25 (Feasible Distance), so this path is still considered a loop B A C D E 98 49

50 DUAL The path through C now fails A examines its topology information, and finds it has no loop free path to E However, it does have a neighbor, and that neighbor might have a loop free path So, it places E in active state and queries D B A C D E 99 DUAL D examines its topology information A Since its best path is not through A, the path it has to E is still valid D sends a reply to this query, indicating it still has a valid loop free path to E Once A receives this reply, it begins using the path through D B C E D

51 Show IP EIGRP Topology show ip eigrp topology displays a list of successors and feasible successors for all destinations known by eigrp A k 128k.2.2 B C k k E k.2 D.1 56k RtrA#show ip eigrp topology IP-EIGRP Topology Table for AS(1)/ID( )..snip.. P /24, 1 successors, FD is via ( / ), Serial1/0 via ( / ), Serial1/1 Feasible Distance Successor Feasible Successor Computed Distance Reported Distance 101 Show IP EIGRP Topology <network><mask> RtrA#show ip eigrp topology IP-EIGRP topology entry for /24 State is Passive, Query origin flag is 1, 1 Successor(s), FD is Routing Descriptor Blocks: (Serial1/0), from , Send flag is 0x0 Composite metric is ( / ), Route is Internal Vector metric: Minimum bandwidth is 128 Kbit Total delay is microseconds Reliability is 255/255 Load is 1/255 Minimum MTU is 1500 Hop count is (Serial1/1), from , Send flag is 0x0 Composite metric is ( / ), Route is Internal Vector metric: Minimum bandwidth is 56 Kbit Total delay is microseconds Reliability is 255/255 Load is 1/255.1 Minimum MTU is 1500 A Hop count is show ip eigrp topology <network><mask> displays detailed information for all paths received for that destination k 128k.2.2 B C k k E k.2 D.1 56k

52 Show IP EIGRP Topology <network><mask> Showing the topology table entry for an external route shows additional information about the route RtrA#show ip eigrp topology IP-EIGRP topology entry for /24 State is Passive, Query origin flag is 1, 1 Successor(s), FD is Routing Descriptor Blocks: (Serial1/2), from , Send flag is 0x0 Composite metric is ( / ), Route is External Vector metric: Minimum bandwidth is 56 Kbit Total delay is microseconds Reliability is 255/255 Load is 1/255 Minimum MTU is 1500 Hop count is 1 External data: Originating router is AS number of route is 0 External protocol is Static, external metric is 0 Administrator tag is 0 (0x ) A k k 56k B C D k k 56k Static Route to /24 Is Redistributed into EIGRP E 103 EIGRP Routing Update Contents 4 Bytes Version Opcode Checksum Flags Sequence Ack Autonomous System Type/Length/Value Section 0x0101 (IP Update) TLV Length Next Hop Forwarding Address Delay Bandwidth MTU Hop Count Reliability Load Reserved Subnet Msk IP Prefix (Variable size) Common Header for all EIGRP Packets Internal Route Update information duplicated for each route

53 Advantages of EIGRP Fast convergence and loop free (DUAL algorithm) Easy configuration Less network design constraints than other routing protocol Incremental update Classless routing (VLSM and discontiguous network support) Utilize link bandwidth and delay for flexible metrics control Unequal cost paths load balancing 105 LINK STATE ROUTING PROTOCOLS

54 Link State Overview In a link-state protocol, the network can be viewed as a jigsaw puzzle Each jigsaw piece holds one router Each router creates a packet which represents its own jigsaw piece These packets flood everywhere Use SPF to put the pieces together 107 Jigsaw Analogy LSP for Rtr A to B to A LSP for Rtr B to E to C to E to A to B to D to A LSP for Rtr E to E to C to D to E to B to C LSP for Rtr D LSP for Rtr C to D

55 Link State Overview In link state protocols, each router advertises the state of its links to every other router in the network A B 10 E determines it is connected to: C D /24, metric 10 C, metric D, metric 10 E advertises this information to C and D E / Link State Overview C and D record this information, and retransmit it to B A 10 This process of recording and retransmitting is called flooding Because information is flooded within a link state network, every router in the network should have the same information about what the network looks like (the same database LSDB) C B E /24 D

56 All Routers Have the Same LSDB Router A s LSDB LSPB LSPA LSPE LSPD LSPC Router C s LSDB LSPB LSPA Router E s LSDB LSPB LSPA LSPE LSPD LSPC Router B s LSDB LSPB LSPA LSPE LSPD LSPC Router D s LSDB LSPB LSPA LSPE LSPE LSPC LSPD LSPC LSPD 111 Link State Overview Each router in the network then uses this information to build a tree which describes the shortest path to each destination in the network A B best path to /24 10 The Dijkstra shortest path first (SPF) algorithm is used to build this tree C D E /

57 Link State OPSF and IS-IS both implement link state routing using the Dijkstra shortest path first algorithm 113 CONCEPTS IN LINK STATE PROTOCOLS Presentation_ID 2003, 2004 Cisco Systems, Inc. All rights reserved

58 Why Is It Called a Link State Protocol? Link State Protocols relay specific link characteristics and state information Only changes or updates are sent across the network Each router uses that information to build a routing table on its own There are two common Link State Routing Protocols OSPF Open Shortest Path First ISIS Intermediate System to Intermediate System 115 Metrics Each link (router interface) is assigned a cost Metric = summation of costs along a path Numeric value that can be administrator assigned, or calculated using link characteristics information More flexible than hops used in Distance Vector routing protocols

59 Topology/Link State Database The Link State Database (LSDB) contains information regarding all links and routers within a logical area A router has a separate LSDB for each area to which it belongs Routers are collected into logical groups called areas All routers belonging to the same area have identical databases for that area A calculation is performed separately for each area 117 Adjacencies Routers participating in a Link State Protocol are uniquely identified throughout the network with a router ID (some form of address) Link state protocol routers discover their adjacent neighbors with the help of Hello protocol Once discovered neighbors form a relationship to exchange/synchronize LSDB information between them

60 Building the Database Hello packets discover neighbors Once neighbors are discovered LSDB information is exchanged A Hello, I m B Hello, I m A Let s exchange information OK I know about these links I know about these links B 119 Building the Database: Hello Protocol Hello packets are used to acquire and maintain neighbor adjacencies Hello packets ensure that neighbors belong to the same logical subnet, same area, or contain the correct authentication information Timers are used to detect when neighbors are no longer present

61 Synchronizing the Database Routers exchange lists describing the information they know Routers request specific information from their neighbor about things they do not know Reception of new information is reliably acknowledged 121 Logical Hierarchy Link State Protocols deploy a logical hierarchy in their design Typically consists of two levels Usually consists of the concept of a backbone level and another sub-level OSPF: Backbone area (area 0), regular areas IS-IS: L2 Backbone areas, L1 areas Primary goal of area boundaries is to limit the flooding of topology information while allowing the propagation of reachability information Enables scalability by summarizing and abstraction, thereby reducing information from lower level areas into the higher level area

62 Not Summarized: Specific Links Backbone Area Only summaries are advertised out of an area Link-state changes do not propagate between areas Reachability information is propagated through ABR Flooding Information that changes or is learned from a neighbor is flooded across a logical network area This is done to maintain consistency of the LSDB across all routers Flooding of Link State Packets (LSP) are contained within an area

63 Flooding Scope OSPF Intra-area, inter-area, external IS-IS Area, backbone 125 Link State Protocols Rely on the Dijkstra Shortest Path First (SPF) algorithm to calculate path tree Maintain a database of network information for a complete picture of the network Form adjacencies between neighboring devices and flood to exchange database information Utilize a hierarchal design to achieve scalability

64 Hiding Topology In link state protocols, we find places we can declare the edge between two areas or routing domains, and block out topology information at that edge A Area Border C B C can reach /24, and I m connected to C! This is called an Area Border Router in OSPF, and an L1/L2 edge in IS-IS E D F G / Hiding Topology Hiding topology information also hides information about changes in the topology If A, C and D can still reach /24 when the G to F link fails, why do they need to know about the failure? A Area Border C D B C can reach /24, and I m connected to C! Hiding the topology between areas is the only way to scale Link State Protocols E F G /

65 Hiding Topology Topology information is naturally hidden in distance vector protocols, beyond the next hop C and D only advertise that they can reach /24, not that they are connected to D, which is then connected to /24 B A D C / Hiding Topology When there are multiple parallel links, distance vector protocols will also overload on information, so we still need to consider hiding topology information How many times is it really necessary for B to tell A it can reach /24? A B /

66 Hiding Topology Hierarchy provides points at which complex sections of the network can be separated from each other In other words, hierarchy provides choke points where topology information can be summarized Choke Points 131 THE DIJKSTRA ALGORITHM Presentation_ID 2003, 2004 Cisco Systems, Inc. All rights reserved

67 Dijkstra Shortest Path First (SPF) Algorithm Dijkstra is a path finding algorithm Will find the shortest path from A to B given intermediate path and cost information One of many path finding algorithms: Dijkstra, best path, A*, etc 133 Dijkstra Shortest Path First (SPF) Algorithm Link state database Created with Link State Packets (LSPs) from each router TENT database Tentative triples (ID, path cost, direction)

68 Dijkstra (SPF) Overview PATH database Best path triples (ID, path cost, direction) Forwarding database A.k.a. the routing table 135 Shortest Path First Algorithm: TWCC Two Way Connectivity Check (TWCC) When a new adjacency is advertised, the calculating router must have BOTH LSPs (LSPs from both ends of the adjacency) in order for the adjacency to be considered during SPF In order for a node to be moved into TENT, it has to report an adjacency to its parent In order for all other routers to process this change (Bad News) only one LSP is needed TWCC will fail during SPF anyway Bad News requires one LSP

69 Shortest Path First Algorithm: TWCC G E LSP-F F LSP-C C D In order to compute a new SPT, we only need one LSP (either LSP-C or LSP-D. TWCC will fail anyway during SPF and C-F link will not be used Link down! Both F and C will generate a new LSP B A 137 An Example Network H F G 12 E 8 B C D S2 S3 S1 A S

70 The Link-State Database LSP H LSP F IS: 4 E IS: 2 C IS: 5 G IS: 5 G LSP E IS: 8 B IS: 4 F LSP G IS: 5 F IS: 5 F IS: 12 D IS: 12 C LSP D IS: 3 A IS: 3 C IS: 12 G LSP B IS: 3 A IS: 4 A IS: 3 C IS: 8 E LSP C IS: 7 A IS: 3 D IS: 3 B IS: 2 F IS: 12 G LSP A IS: 3 B IS: 4 B IS: 7 C IS: 3 D 139 Adjacency Database (Router-a) Neighbor Interface Cost B serial0 3 B serial1 4 C serial2 7 D serial

71 Shortest Path First Example Initial situation TENT: A PATHS: empty Unknown: B C D E F G H 141 Shortest Path First Example H 3 5 G First iteration Move A (ourself) to PATHS with cost 0 Move neighbors of A to TENT: B, C, D Note that TWCC succeeds since B, C, D reports an adjacency to A Find first-hop info in adjacency database TENT: D cost 3 via S3, C cost 7 via S2, B cost 3 via S0 E F B Unknown C S2 S3 S1 A S0 D PATHS: Unknown: A E F G H TENT PATHS

72 Shortest Path First Example H F G 12 Second iteration Move D to PATHS with cost 3 and insert in RIB its prefixes Move neighbors of D to TENT: G, found better path to C, ignore A TENT: C cost 6 via S3, B cost 3 via S0, G cost 15 via S3 PATHS: A, D cost 3 via S3 Unknown: E F H E 8 3 B Unknown TENT 4 3 C 3 D 7 3 S2 S3 S1 A S0 PATHS 143 Shortest Path First Example Third iteration Move B to PATHS with cost 3 and insert in RIB its prefixes Move neighbors of B to TENT: E, found equal cost path to C, ignore A TENT: C cost 6 via S3 and S0, G cost 15 via S3, E cost 11 via S0 PATHS: A, D cost 3 via S3, B cost 3 via S0 Unknown: F H E H F B 2 3 Unknown TENT PATHS C 3 G D 7 3 S2 S3 S1 A S

73 Shortest Path First Example Fourth iteration Move C to PATHS with cost 6 and insert in RIB its prefixes Move neighbors of C to TENT: F, ignore G, ignore B, ignore D, ignore A TENT: G cost 15 via S3, E cost 11 via S0, F cost 8 via S3 and S0 PATHS: A, D cost 3 via S3, B cost 3 via S0, C cost 6 via S3 and S0 Unknown: H E H F B 2 3 Unknown TENT PATHS C S0 3 G D 7 3 S2 S3 S1 A 145 Shortest Path First Example Fifth iteration H F G 12 Move F to PATHS with cost 8 and insert in RIB its prefixes Move neighbors of F to TENT: Better path to G, ignore C, ignore E TENT: G cost 13 via S3 and S0, E cost 11 via S0 PATHS: A, D cost 3 via S3, B cost 3 via S0, C cost 6 via S3 and S0, F cost 8 via S3 and S0 Unknown: H E 8 3 B Unknown TENT PATHS 4 3 C S2 S3 S1 A S0 D

74 Shortest Path First Example Sixth iteration H F G 12 Move E to PATHS with cost 11 and insert in RIB its prefixes Move neighbors of E to TENT: Ignore B, ignore F TENT: G cost 13 via S3 and S0 PATHS: A, D cost 3 via S3, B cost 3 via S0, C cost 6 via S3 and S0, F cost 8 via S3 and S0, E cost 11 via S0 Unknown: H E 8 B Unknown TENT PATHS C 3 D 7 3 S2 S3 S1 A S0 147 Shortest Path First Example Seventh iteration Move G to PATHS with cost 13 and insert in RIB its prefixes Move neighbors of E to TENT: Ignore F, ignore C, ignore D TENT: EMPTY!!! PATHS: A, D cost 3 via S3, B cost 3 via S0, C cost 6 via S3 and S0, F cost 8 via S3 and S0, E cost 11 via S0 G cost 13 via S3 and S0 Unknown: H E H F B 2 3 Unknown TENT G C 3 D 7 3 S2 S3 S1 A S0 PATHS

75 Shortest Path First Example TENT is empty => iteration stops Result: A, D cost 3 via S3, B cost 3 via S0, C cost 6 via S3 and S0, F cost 8 via S3 and S0, E cost 11 via S0, G cost 13 via S3 We obviously didn t find a path to H 149 Shortest Path First Example The Shortest Path Tree Not all links are used Still if a change occurs in one of these unused links, the full SPF algorithm need to be re-run F Cost: 8, NH: D, B G Cost: 13, NH: D E Cost: 11, NH: B C Cost: 6, NH: D, B D Cost: 3, NH: D B Cost: 3, NH: B S1 S2 S3 S0 A Cost: 0, NH:

76 LINK STATE PROTOCOL COMPARISON Presentation_ID 2003, 2004 Cisco Systems, Inc. All rights reserved. 151 IS-IS and OSPF: Similarities IS-IS and OSPF are Interior Gateway Protocols (IGP) They distribute routing information between routers belonging to a single Autonomous System (AS) With support for: Classless Inter-Domain Routing (CIDR) Variable Subnet Length Masking (VLSM) Authentication Multipath IP unnumbered links

77 Terminology in IP LSPs Host Router Link Packet OSPF IS-IS End System (ES) Intermediate System (IS) Circuit Protocol Data Unit (PDU) Designated Router (DR) Backup DR (BDR) Link-State Advertisement (LSA) Hello Packet Designated IS (DIS) N/A (no BDIS is used Link-State PDU (LSP) IIH PDU Database Description (DBD) Complete Sequence Number PDU (CSNP) 153 Terminology (Cont.) OSPF Area Non Backbone Area Backbone Area IS-IS Sub domain (area) Level-1 Area Level-2 Sub domain (backbone) Area Border Router (ABR) L1-L2 Router Autonomous System Boundary Router (ASBR) Any IS

78 Media Handling OSPF IS-IS Point-to-point Broadcast (LAN Segments) Non-Broadcast Point-to-point Broadcast *Non-Broadcast Point-to-Multipoint *Non-Broadcast is supported on ISIS, but is discouraged 155 LSDB Management OSPF OSPF Counts up to MaxAge (60 minutes) It is not configurable Thus, every LSA needs to be refreshed every LSRefreshTime period (30 minutes) Refresh is a fixed constant IS-IS IS-IS counts down to 0 LSPs get flushed when age reaches 0 LSPs are originated with a configurable non-zero value Thus refreshment interval is configurable (default is 15 minutes)

79 OSPF AND IS-IS TRANSPORT Presentation_ID 2003, 2004 Cisco Systems, Inc. All rights reserved. 157 OSPF and IS-IS: Transport OSPF uses IP Protocol 89 as transport Data Link Header IP Header OSPF Header OSPF Data IS-IS is directly encapsulated in Layer 2 Data Link Header IS-IS Header IS-IS Data

80 ISIS: Transport NSAP: Network Service Access Point An NSAP mainly consists of 3 parts: Area Address System ID PN-ID Variable Length 6 Bytes (fixed) 1 Byte Example: Area. System ID. Pseudonode ID 159 OSPF: Router ID Makes use of IPv4 addressing Routers are identified by a unique ID Router ID can be assigned automatically or manually configured router-id <ip address> RID: Highest IP address from any loopbacks RID: If no loopback exists, highest IP address configured to a physical interface Router ID used to build the SPT

81 PACKET TYPES Presentation_ID 2003, 2004 Cisco Systems, Inc. All rights reserved. 161 OSPF Protocol Packets Share a common protocol header Routing protocol packets are sent with type of service (TOS) of 0 Five types of OSPF routing protocol packets Hello packet type 1 Database description packet type 2 Link-state request packet type 3 Link-state update packet type 4 Link-state acknowledgement packet type

82 OSPF Hello Packet Hello [Packet Type 1] Establishes neighbor relationships Elects the designated router Describes the optional capabilities Send to multicast address of (all SPF routers) Router priority decides the designated router Network Mask Hello Interval Options Rtr Pri Router Dead Interval Designated Router Backup Designated Router Neighbor 163 OSPF DBD Packet Database Description Packet [Packet Type 2] Exchanged when adjacency is initialized Describes the contents of the link-state database Multiple packets are used to describe the database One router is designated as master and the other as slave Processing of DBD packet depends on neighbor state If the interface MTU field in the DBD packet indicates a larger IP datagram size than the router can accept, the packet is rejected Interface MTU Options DD sequence number An LSA Header 00000I MMS

83 OSPF Link State Request Packet Link State Request [Packet Type 3] If parts of database are missing If parts of database are out of date A precise piece of the database is requested via the sequence #, LS checksum and LS age LS type Link State ID Advertising Router 165 OSPF Link State Update Link State Update [Packet Type 4] Implements flooding of LSA Carries collection of LSA one hop further from their origin Several LSAs are included in a single packet Update packets are multicast on physical networks that support multicast # LSA s LSA s

84 OSPF Link State Acknowledgement Link State Acknowledgement [Packet Type 5] Used to acknowledge each LSA Multiple LSAs can be acknowledge in a single ack Ack can be implicitly accomplished by sending a link state update packet Links state ack are sent to multicast addresses If the state of the router is DR/BDR then ack is sent to all SPF routers If the state is DR other then it is sent to all DR routers LS Age Options LS Type Link State ID Advertising Router LS Sequence number LS checksum Length 167 OSPF: Types of LSAs Five distinct type of LSAs Router link LSA type 1 Network link LSA type 2 Network summary LSA type 3 ASBR summary LSA type 4 External LSA type 5 NSSA external LSA type

85 OSPF: Different Type of LSAs Router LSA [Type 1] Describes the state and cost of the router s links to the area All of the router s links in an area must be described in a single LSA Flooded throughout the particular area and no more Router indicates whether it is an ASBR, ABR, or end point of virtual link A OSPF Router with ID ( ) (Process ID 10) Router Link States (Area 0) LS age: 144 Options: (No TOS-capability, DC) LS Type: Router Links Link State ID: Advertising Router: LS Seq Number: Checksum: 0x6B12 Length: 48 Number of Links: 2 Link connected to: another Router (point-to-point) (Link ID) Neighboring Router ID: (Link Data) Router Interface address: Number of TOS metrics: 0 TOS 0 Metrics: 64 B Link connected to: a Stub Network (Link ID) Network/subnet number: (Link Data) Network Mask: Number of TOS metrics: 0 TOS 0 Metrics: OSPF: Different Type of LSAs Network LSA [Type 2] Generated for every transit broadcast and NBMA network Describes all the routers attached to the network Only the designated router originates this LSA Flooded throughout the area and no more A OSPF Router with ID ( ) (Process ID 10) Net Link States (Area 0) Routing Bit Set on this LSA LS age: 23 Options: (No TOS-capability, DC) LS Type: Network Links Link State ID: (address of Designated Router) Advertising Router: LS Seq Number: Checksum: 0x2A70 Length: 32 Network Mask: /24 Attached Router: Attached Router: B

86 OSPF: Different Type of LSAs Summary LSA [Type 3 and 4] Describes the destination outside the area but still in the AS Flooded throughout a single area Originated by an ABR Only intra-area routes are advertised into the backbone Type 4 is the information about the ASBR A OSPF Router with ID ( ) (Process ID 10) Summary Net Link States (Area 0) Routing Bit Set on this LSA LS age: 129 Options: (No TOS-capability, DC, Upward) LS Type: Summary Links(Network) Link State ID: (summary Network Number) Advertising Router: LS Seq Number: Checksum: 0x65FA Length: 28 Network Mask: /24 TOS: 0 Metric: 10 B Area 1 C 171 OSPF: Different Type of LSAs External LSA [Type 5] Defines routes to destination external to the AS Default route is also sent as external Two types of external LSA: E1: Consider the total cost up to the external destination E2: Considers only the cost of the outgoing interface to the external destination A B External Routes OSPF Router with ID ( ) (Process ID 10) Type-5 AS External Link States Routing Bit Set on this LSA LS age: 26 Options: (No TOS-capability, DC) LS Type: AS External Link Link State ID: (External Network Number ) Advertising Router: LS Seq Number: Checksum: 0x6BA2 Length: 36 Network Mask: /32 Metric Type: 2 (Larger than any link state path) TOS: 0 Metric: 13 Forward Address: External Route Tag:

87 OSPF: Different Type of LSAs NSSA External [Type 7] Defines a more flexible, hybrid method for Stub areas Allows injection of external routes into the Stub Area Redistribution into an NSSA area creates a special type of link LSA Type 7 LSA Type 7 can only exist in an NSSA area NSSA ABR translates it into a type 5 LSA, which gets propagated into the OSPF domain A OSPF Router with ID ( ) (Process ID 10) Type-7 AS External Link States (Area 1) Routing Bit Set on this LSA LS age: 210 Options: (No TOS-capability, Type 7/5 translation, DC) LS Type: AS External Link Link State ID: (External Network Number ) Advertising Router: LS Seq Number: Checksum: 0x9324 Length: 36 Network Mask: /32 Metric Type: 2 (Larger than any link state path) TOS: 0 Metric: 10 Forward Address: External Route Tag: 0 B NSSA Area C 173 IS-IS PDUs Hello PDUs (IIH, ISH, ESH) Link State Packet (LSP) Non-pseudonode LSP Pseudonode LSPs Complete Sequence Number PDU Partial Sequence Number PDU

88 IS-IS: Node and Pseudonode LSP 2 kinds of Link State PDUs Non-Pseudonodes represent routers Pseudonodes represents LANs (created by the DIS) Level 1 routers create Level 1 LSP Level 2 routers create Level 2 LSP Level 1 2 routers create Level 1 LSP and Level 2 LSP 175 FLOODING Presentation_ID 2003, 2004 Cisco Systems, Inc. All rights reserved

89 Flooding Scope OSPF intra-area, inter-area and external IS-IS area, backbone 177 OSPF: Types of Flooding Flooding on p2p links Each LSA is acknowleged Flooding on LANs DR is responsible (Network LSA) General background flooding Refreshing LSAs

90 OSPF: LSA Flooding Type 2 Area 1 Type 3/4/5 ABR Type 3 LSA Types 1. Router 2. Network 3. Summary 4. ASBR Summary 5. External Area 0 Type 3/4/5 ABR Type 3 Area 2 ASBR Type 5 RIP Domain 179 OSPF: LSA Flooding Router-LSA Is originated by Router A for the area that it belongs to It describes the Router A links to the area It is only flooded in the specific area and no further A E C Area 0 D G Area 2 Area 1 B F /

91 OSPF: LSA Flooding Network LSA Is Originated by Router G for the NBMA Network It describes all the routers that are attached to the network The DR on the LAN will originate the LSA E D G F / OSPF: LSA Flooding Summary LSA Is generated by Router D which is an area border router (ABR) The LSA is flooded throughout a single area only E C Area 0 D Area 1 F AS External routes are never advertised in the Summary LSAs G /

92 OSPF: LSA Flooding External LSA Router G will flood the External LSAs into the OSPF domain Describes the routes to the destinations external to the AS External LSAs are the only LSAs that get flooded into the entire OSPF Domain [not Stub] C Area 0 D E Area 1 G /24 External Redistribution F 183 OSPF: Function of DR/BDR DR Flood Link change Link change BDR

93 OSPF: Designated Routers Reduce OSPF traffic on multi-access links Store and distribute neighbor s LSDBs Backup DR for redundancy 185 IS-IS: Types of Flooding IS-IS has only two type of Flooding: on p2p links with positive acks each LSP is acknowleged with a PSNP on LANs with negative acks DIS multicasts a full list of LSP descriptions in a CSNP packet every 10 seconds Re-transmission requests are done via PSNP General background flooding Refreshing LSPs

94 IS-IS: SRM and SSN Bits SRM [Send Routing Message] SRM Bits: set per interface when an IS has to flood this LSP SSN [Send Sequence Number] SSN Bits: set per interface when an IS has to send a PSNP for this LSP 187 IS-IS: p2p Flooding Once the adjacency is established between the Routers, they send a CSNP packet, which describe the complete set of LSPs Any missing LSPs are sent by the Router, if not present in the received CSNP Missing LSPs may be requested through PSNP SRM bit is set when LSP is flooded and only cleared once an ack is received A C B

95 IS-IS: Flooding on LAN The LAN flooding is reliable due to the Designated Intermediate System (DIS) DIS is like DR in OSPF DIS election is based on priority Breaking-tie is the highest SNPA (MAC address) DIS election is deterministic G /24 H I J 189 IS-IS: Flooding on LAN DIS has two tasks Creating and updating the Pseudonode LSP Conducting the flooding over the LAN A DIS is elected for each LAN DIS broadcasts CSNPs every 10 seconds No Backup DIS in ISIS [not necessary, no LSPD resync] G /24 PSN H I J

96 ADJACENCY ESTABLISHMENT 191 Discovering Neighbors: Hello Protocol Both IS-IS and OSPF use a Hello mechanism for discovering neighbors and maintaining adjacencies IS-IS uses the following defaults: Hello Interval: 10 seconds Hold Time: 30 seconds OSPF uses the following defaults: Hello Interval Dead Interval 10 seconds 40 seconds

97 The Hello Protocol Responsible for establishing and maintaining neighbour relationships Elects designated router on multi-access networks B Hello A D Hello Area 1 Hello Hello C /24 E 193 OSPF: Hello Packets Router priority Hello interval Router dead interval Network mask Options: TOS-bit List of neighbours B Hello A Hello Area 1 Hello Hello D /24 C Multicast to E

98 IS-IS: Hello Packets IIHs are between routers Exchanged by IS to form adjacencies Point-to-point IIH Level 1 LAN IIH Level 2 LAN IIH 195 IS-IS: L1 Adjacency Router A and Router B will form L1 Adjacency only when both the routers are in the same Level-1 Area Both Routers will not form L2 Adjacency if they belong to different areas If IS-IS is running for IP, then the IP-address and the subnet associated with the link will also be checked during the adjacency formation L1-Nei. Because the area is the same A B

99 IS-IS: L2 Adjacency Router A and B will only form L2 adjacency L1 adjacency will not form as both the routers belong to different areas L2 Neioborship established A B IS-IS: Adjacency Only two active states: INIT and UP Adjacency makes synchronisation of LSDB possible: On p2p through the ack ing of the LSP s On broadcast through frequently send CSNP packets

100 OSPF: Adjacency States 2-way Router sees itself in other Hello packets DR selected from neighbours in state 2-way or greater 2-Way DR BDR 199 OSPF: Adjacency States Full Routers are fully adjacent Databases synchronised Relationship to DR and BDR Full DR BDR

101 AREAS LEVELS 201 Hierarchy Levels Both IS-IS and OSPF Have a 2 Layer Hierarchy OSPF: The backbone (Area 0) The areas IS-IS: The backbone (level-2 domain) The areas (level-1)

102 OSPF Areas Area 0 is the Backbone Area Both Router C and Router D are in area A C Area 2 B Backbone area contiguous Area 0 All other areas must be connected to the backbone E D Area 1 F G / OSPF Areas Routers C and D are also acting as Area Border Routers (ABR) They are connected to both Area 0 and other Area Router C: Area 0 and Area 2 A E C Area 0 D Area 2 Area 1 B F Router D: Area 0 and Area 1 G /

103 OSPF Areas Router G will be acting as an Autonomous System Boundary Router (ASBR) as it is connected to outside the domain A C Area 0 Area 2 B D E Area 1 F G To other AS / OSPF Areas Intra Area Route All routes inside an area. They are the routes generated (as an example) by Router A and Router B Inter Area Route The routes advertised from one area to another by an Area Border Router (ABR) [Router C and D] External Route Routes imported into OSPF from other protocol or static routes [Router G] A C Intra-Area Routes Area 0 E External Route D G To other AS Area 2 Area 1 Inter-Area Route B F /

104 OSPF Areas Stub Areas Router A is in a Stub Area It does not accept routes belonging to external autonomous systems (AS); however, it will have interarea and intra-area routes A Stub Area C Area 0 Area 2 B Router A will use the default route injected into the area by the ABR [Router C] D 207 OSPF Areas Totally Stubby When Router A is configured for Totally Stubby Area we do not allow routes other than intra-area and the default routes to be propagated within the area A Totally Stub Area C Area 2 B The ABR [Router C] injects a default route into the area and all the routers belonging to this area use the default route to send any traffic outside the area Area 0 D

105 OSPF Areas Totally Not So Stub When Router A is configured for NSSA, then it generates an NSSA external link-state advertisement (LSA) (Type-7) which can be flooded throughout the NSSA area These Type-7 LSAs are converted into Type-5 LSAs at the NSSA ABR and flooded throughout the OSPF domain External Route To other AS A Totally Stub Area C Area 0 D Area 2 B 209 IS-IS: Hierarchy Levels IS-IS presently has a 2 layer hierarchy The backbone (level-2) The areas (level-1) An IS (router) can be either: Level-1 router (intra-area routing) Level-2 router (inter-area routing) Level-1-2 router (intra and inter-area routing)

106 IS-IS: Level 1 Routers Level 1 Routers are those which belong in the same area Routers: A, B and C will be in Level 1 if they share the same area The Level 1 Database will have only intra-area information A L1 L1 Adj L1 B Level 1 only routers install a default route to the closest Level-1-2 router in the area L1L2 C 211 IS-IS: Level 2 Routers Level 2 is synonymous with Area 0 in OSPF Routers C, D, E and F are Level 2 Routers L2 routers know about other area L2 area addresses and L2 routers When doing OSI routing, the L2 ISs must know their own area Therefore never use L2 only on OSI routers A E L2 C D Level-1 L2 L2 B Level-2 F L2 L2 only is possible when doing just IP G Level

107 IS-IS: Level 2 Routers By default, Router D will form L1 adjacency with Routers: C, E and F as they are all in the same area But, it needs to have an L2 database as well, since Router D is a Transit node A L1 C Level-1 L1L2 L1 Level-2 B So, for transit traffic requires routers inside the area to know about other areas Hence Router D has to have L1-L2 or L2 Adjacency L1-L2 as a default may not be optimal if IS-IS is running for IP E L1L2 D G L1L2 / L1 L1 L1L2 Level-1 F 213 IS-IS: Level 2 Routers For Router G to reach either Router A or B, there is a backbone connectivity via Router E and C But if the link between E and C goes away, then the backbone is partitioned This is not desirable and will result in loss of connectivity in the domain A E L1 L1L2 C G Level-1 L1L2 L1 L1 L1 B Level-2 D F L1L2 Level

108 IS-IS: Level-2 Routers Backbone MUST be L2 contiguous Router D should have either L1-L2 or L2 adjacency to act as a transit router So that there is a connectivity from end-to-end A E L1 L1L2 C Level-1 L1L2 L1L2 L1 B Level-2 D F L1L2 G L1 Level LINK STATE VS. DISTANCE VECTOR

109 Link-State vs. Distance Vector Protocols Link-State router tells ALL other routers about ONLY its neighbors and links Distance Vector router tells ONLY neighbors about ALL routes 217 Link-State vs. Distance Vector Protocols In Link-State protocols Update and Decision processes are independent With Distance Vector protocols a router can t send out new vectors until it has processed them Link-State are fast loop-free convergence protocols

110 Advantages of a Link State Protocol Uses metrics (costs) to calculate path Typically faster convergence than distance vector routing protocols Typically more scalable due to hierarchical nature 219 Disadvantages of a Link State Protocol Requires more planning and understanding to deploy

111 Cisco Press Publications ISBN: ISBN: Cisco Press Publications ISBN: ISBN:

112 Other Related Networkers Sessions: Advanced Level Deploying Interior Gateway Routing Protocols Troubleshooting Link State Protocols 223 Complete Your Online Session Evaluation! WHAT: WHY: Complete an online session evaluation and your name will be entered into a daily drawing Win fabulous prizes! Give us your feedback! WHERE: Go to the Internet stations located throughout the Convention Center HOW: Winners will be posted on the onsite Networkers Website; four winners per day

113 9906_06_2004_c