EIGRP An In-Depth Look At The Protocol

Transcription

1

2 EIGRP n In-Depth Look t The Protocol RKIPM-2444 Donnie Savage Donald Slice 2

3 The History and Evolution of EIGRP EIGRP introduction and Feature Roadmap DUL published to IEEE/CM 1993 EIGRP Shipped 1994 Interop SI Rewrite 1999 Hub and Spoke PE/CE Support 2001 Neighbor Reliability NSF/SSO 2003 Route-Maps 3-Way Handshake rd Party Next-hop SNMP FD Support Pix Firewall IPv6 Support FD Prefix Limits Site of Origin 2006 DMVPN SRP PfR Manet Plugin Support 2007 Cross Licensing Service Family 2008 Code Harding Unified CLI Stub Leaking Summary Leaking 2009 Summary Metric EVN Support IPv6 VRF HMC SH2 uthentication Peer Groups Remote Peers 2011 Wide Metrics DMVPN Phase 1 IPX Disabled 2012 DMVPN Phase 2 IPv6 FD/NSF/MIS Route-Tag Enhancements IP FRR 2013 EIGRP OTP nd it s going to get even better in 2013! 3

4 EIGRP Flexible Support for Hierarchy How many levels of Hierarchy do you need?? Two layer? Three layer? More EIGRP is not limited! Depth of the hierarchy doesn t alter the way EIGRP is deployed; there are no hard edges Summarize High Degree of Complexity Core Distribution ccess Rule of Thumb: alance simplicity, optimal routing, and functional separation High Degree of Complexity 4

5 Milliseconds EIGRP Convergence Comparative Data We can sort typical convergence times into three groups IPv4 IGP Convergence Data EIGRP with feasible successors IS-IS with tuned timers OSPF with tuned timers EIGRP without feasible successors OSPF with default timers IS-IS with default timers Route Generator C Routes D 5

6 Convergence Time EIGRP Peer Scaling Initial convergence testing was done with 400 peers with 10,000 prefixes to each peer Measure initial bring up convergence until all peers are established and queues empty EIGRP DMVPN Phase 0 (prior to 12.4(7)) EIGRP DMVPN Phase I (12.4(7) and later) EIGRP DMVPN Phase II (CSCei03733) Currently EIGRP offers support for over 3500 peers Work is in progress to considerably improve this peer count min 11 min 3 min Phase 0 Phase I Phase II 6

7 The History and Evolution of EIGRP With proper network design EIGRP is the: Easiest to configure Easiest to deploy Fastest converging Highest scaling Given it s such a great protocol, why isn t every one running it? Its not 1995 any more Standards and Multi-Vendor are key So let s fix that! 7

8 EIGRP Moving into the Future EIGRP Information Draft published to IETF nnounced at Cisco Live London Competitive Landscape; Currently there are at least 4 known companies shipping -EIGRP in sia and Europe today. In talks with major US based Venders IPv6 is offering a green-field deployment to customers who want "standards based solutions. Pressure from public/government sectors who have mandates to use Open solutions when available Removes the "standards" argument now allows customers to use the technology that best fits their needs. Development of new features and better scaling are in progress Cisco is committed to continue offering best of breed 2013 Open EIGRP: draft-savage-eigrp-00 8

9 The Diffusing Update lgorithm (DUL)

10 The Diffusing Update lgorithm (DUL) Definition of Key Terms Reported Distance (RD) Reported Distance is the total distance along a path to a destination network as advertised by an upstream peer Feasible Distance (FD) Feasible Distance is the lowest distance to a particular destination; that is, its the Reported Distance plus link cost to reach the upstream peer Feasibility Condition If a router s Reported Distance is less than our Feasible Distance, then this router is a loop-free path to this destination and meets the Feasibility Condition Feasible Successor router is a Feasible Successor if it satisfies the Feasibility Condition for a particular destination Successor router is a Successor if it satisfies the Feasibility Condition ND it provides the lowest distance to that destination ctive and Passive State route is in the Passive state when it has a successor for the destination. The router goes to ctive state when current successor no longer satisfies the Feasibility Condition and there are no Feasible Successors identified for that destination 10

11 The Diffusing Update lgorithm (DUL) Definition of Key Terms R# show eigrp address-family ipv4 topology EIGRP-IPv4 Topology Table for S(1)/ID( ) Codes: P - Passive, - ctive, U - Update, Q - Query, R Reply, r - reply Status, s - sia Status P /24, 1 successors, FD is via ( / ), FastEthernet1/0 via ( / ), FastEthernet1/1 Feasible Distance Successor State Computed Distance Reported Distance Feasible Successor 11

12 The Diffusing Update lgorithm (DUL) How DUL works asic idea - if your metric to the destination is less than mine, the path through you can not be a loop N Consider the following network For simplicity, all links cost (1) Each router has the following Feasible Distance to reach N: : Successor = N, FD=1 : Successor =, FD=2 C: Successor =, FD=2 D: Successor =, FS = C, FD= C D Note: From Router-D point of view, both Router- and Router-C are equal cost. ut Router-D must pick one as a successor, the other will be the FS. Which one Router-D picks will be dependent on the order Router-D received the information 12

13 The Diffusing Update lgorithm (DUL) How DUL works If the link between and fails; sends a query informing its peers that it has lost its successor D receives the query and determines if it has any other feasible successors. If it does not, it has to start a route computation and enter the active state N 1 1 However in this case, C is a feasible successor because its cost (2) is less than than D's current cost (3) to destination N D will switch to C as its new successor C D Note and C did not participate because they were unaffected by the change. 13

14 The Diffusing Update lgorithm (DUL) How DUL works Now let's cause a route computation to occur If the link between and C fails; N C determines that it has lost its successor and has no other feasible successors D is not considered a feasible successor because its advertised metric (3) is greater than C's current cost (2) to reach destination N 1 1 C sends a query to its only peer D D replies because its successor has not changed When C receives the reply it can choose its new feasible successor D with a cost of (4) to reach destination N C 1 1 D 1 Note and were unaffected by the topology change and D needed to simply reply to C. 14

15 The Diffusing Update lgorithm (DUL) Choosing a Successor/Feasible Successor How does EIGRP determine which routes are loop free? Each of s peers is reporting reachability to E with a cost of 10 C with a cost of 10 D with a cost of 30 These three costs are called the reported distance (RD); the distance each peer is reporting to a given destination C E D 15

16 The Diffusing Update lgorithm (DUL) Choosing a Successor/Feasible Successor t, the total cost to reach E is: 20 through through C 45 through D D The best of these three paths is the path through, with a cost of 20 This is the feasible distance (FD) 10 C E 16

17 The Diffusing Update lgorithm (DUL) Choosing a Successor/Feasible Successor uses these two pieces of information to determine which paths are loop free The best path (FD) is used as a benchmark; all paths with RD s lower than the FD cannot contain loop C 15 D The algorithm may mark some loop free paths as possible loops It is guaranteed never to mark a looped path as loop free 10 E

18 The Diffusing Update lgorithm (DUL) Choosing a Successor/Feasible Successor For example, consider how views the path available path: The path through is the best path (FD), at 20 C can reach E with a cost of 10; 10 (RD) is less than 20 (FD), so this path is loop free D can reach E with a cost of 30; 30 (RD) is not less than 20 (FD), so EIGRP assumes this path could be a loop C E D 18

19 The Diffusing Update lgorithm (DUL) Fail over from Successor to Feasible Successor What happens if the best path fails, through (Successor)? EIGRP will examine the available paths to E Finding a path previously declared loop free (Feasible Successor), it begins using it immediately C now becomes the Successor (best path) C D E 20

20 The Diffusing Update lgorithm (DUL) Fail over from Successor to Feasible Successor now examines its topology information based on the new successor metric to determine if there is an Feasible Successor (FS) available The Reported Distance (RD) through the remaining peer is D, is 30; - 30 (RD) is more than 25 (FD) So this path is still considered a possible loop C D E 21

21 The Diffusing Update lgorithm (DUL) Fail over from Successor to Non Feasible Successor What if the path through C now fails? examines its topology information, and finds it has no loop free path to E However, it does have a peer, and that peer might have a loop free path So, it places E in active state and Queries D C D E 22

22 The Diffusing Update lgorithm (DUL) Fail over from Successor to Non Feasible Successor D receives a query from Router-D and examines its topology information Since its best path is not through, the path it has to E is still valid D D sends a reply to this query, indicating it still has a valid loop free path to E Once receives this reply, it begins using the path through D 10 C E

23 Session and Topology Management

24 Session and Topology Management 5 asic Packet Types Hello/cks Hellos are used for peer discovery/maintenance. They do not require acknowledgment. hello with a non-zero sequence number is also used as an acknowledgment (ack). Hellos are normally multicast and cks are always sent using a unicast address Updates Updates are used to convey reachability of destinations. When a new peer is discovered, update packets are sent so the peer can populate its topology table. In some cases, update packets will be unicast. In other cases, such as a link cost change, updates are multicast. Updates are always transmitted reliably Query/Reply Queries are sent when destinations go into ctive state. Queries are normally multicast to all peers on all interfaces except for the interface to the previous successor. If a receiving peer does not have an alternative path to the destination, it will in turn Query its peers until the query boundary is reached. Once the Query is sent, the router must wait for all the Replies from all peers before it can compute a new successor. Replies are sent containing the answer (which may be a valid metric or infinity/not reachable) and are unicast to the originator of the query. oth queries and replies are transmitted reliably SI-Query/SI-Reply If any peer fails to Reply to a Query, the destination is said to be Stuck In ctive (SI), and the peer may be reset. t ½ the SI time (default 90 seconds) the router will send an SI-Query to the non-replying peer. The peer must send either an SI-Reply indicating the destination is still active, or a Reply. oth SI-Queries and SI-Replies are transmitted reliably 26

25 Session and Topology Management EIGRP Sends both reliable and Unreliable Packet types EIGRP sends both unreliable and reliable packets Unreliable packets are: Hellos cknowledgement Reliable packets are: Updates Queries Replies SI-Queries SI-Replies Reliable packets are sequenced, require an acknowledgement, and are retransmitted up to 16 times if not acknowledged 27

26 Session and Topology Management Neighbor Formation Router must receive an initial Hello from before it will accept reliable packets from him When receives the first Hello from, it places in the pending state transmits a unicast NULL Update which has the initialization (init) bit set, and no routing data While has is in this state, will not send it any Queries or other Updates to Hello Dst= Hello Dst= in Pending Update Dst= Init=1 28

27 Session and Topology Management Neighbor Formation When receives this Update with the init bit set, it sends an Update with the init bit set as well The acknowledgement (ck) for s initial Update is piggybacked onto this Update packet it is never transmitted by itself Update Dst= Init=1 in Pending There is no way for to receive the ck for its initial Update without also receiving s initial unicast Update Update Dst= Init=1 ck=1 29

28 Session and Topology Management Neighbor Formation While waiting on the acknowledgement, Query and Update packets from are ignored If the acknowledgement is never received, will time out, and the process will start over Once the ck for its initial update is received, moves from pending state to up state then begins sending it s full topology information to Update Dst= Init=1 ck=1 Up State Update Dst= w/topology This completes the 3-Way handshake 30

29 Session and Topology Management Neighbor Formation For each route sends to, sends a poison reverse This makes certain the two router s tables are accurate as well as making sure other routers on the interface they share use the right path for each prefix When a router finishes sending its table, it sends an end-of-table indicator Up State Update Dst= w/topology Update Dst= Route metric= Update Dst= Topo EOT=

30 Session and Topology Management Holding Time Expired The holding time expires when an EIGRP packet is not received during the hold time - Hellos are sent every five seconds on most networks - The hold timer for each peer is reset back to the hold time when NY EIGRP packet is received from that peer Why would a router not see EIGRP packets from a peer?? Router may be gone (crashed, powered off, disconnected, etc.)? Link may be overly congested (input/output queue drops, etc.)? Faulty network dropping packets (CRC errors, frame errors, excessive collisions) Packet 32

31 Session and Topology Management Retry Limit Exceeded Reliable packets are re-sent after Retransmit Time Out (RTO) Typically 6 x Smooth Round Trip Time (SRTT) Minimum 100 ms Maximum 5000 ms (five seconds) 16 retransmits takes between 40 and 80 seconds ck If a reliable packet is not acknowledged before 16 retransmissions and the hold timer duration has passed, reinitialize the peer Packet 33

32 Session and Topology Management Sequence Numbers and cks n Update packet transmitted contains a sequence number that is acknowledged by a receipt of a ck packet If the Update or the ck packet is lost on the network, the Update packet will be retransmitted Update SEQ=100, ck=0 In the case of the Query packets, the Query packet also must be acknowledged ( I heard the question, later followed by a Reply packet ( Here is the answer ). Note that both responses are required and perform different functions ck Seq=0, ck=100 Replies also contain sequence numbers and must be acknowledged 34

33 Session and Topology Management Conditional Receive (CR-mode) sends a multicast Update packet, C and D receive the Update and send an acknowledgment s acknowledgments is lost on the network C D efore the retransmission timer expires, has an event that requires it to send a new multicast update on this interface Update 100 detects that has not cked the last packet and enters the Conditional Receive process Dst= CR-mode builds a Hello packet with a SEQUENCE TLV containing s address This special Hello packet is multicasted unreliably out the interface Hello Dst= list=

34 Session and Topology Management Conditional Receive (CR-mode) C and D process the special Hello packet looking for their address in the list. If not found, they put themselves in Conditional Receive (CR-mode) mode ny subsequent reliable packets received on C and D with the CR-flag set are accepted and processed does not put itself in CR-mode because it finds its address in the list Reliable packets received by with the CR-flag must be discarded and not acknowledged Once has sent the CR Update(s), it exits CR-mode will unicast the previous, unacknowledged packets directly to C D Update 101 Dst= CR=1 Exit CR-mode Update 100, 101 Dst= CR=0 36

35 Query or Not to Query

36 Query or Not to Query The ctive Process What happens when route to /24 is lost? No FS, mark route active Set a 3 minute active timer Query all peers () receives s query No FS, mark route active Set 3 minute active timer Query all peers (C) C receives s query Examine local topology table No feasible successors No peers to query! /24 C /24 Gone; No FS ctive timer set Query /24 Gone; No FS ctive timer set Query /24 Gone 38

37 Query or Not to Query The ctive Process C has no alternate path to /24 Remove from local tables Reply to querying peers receives C s reply No outstanding queries Remove from local tables Reply to querying peers receives s reply No outstanding queries Remove from local tables /24 C C /24 Gone; No FS Remove /24 Query Reply /24 Gone; No FS Remove /24 Query Reply /24 Gone Remove /24 39

38 Query or Not to Query The ctive Process If C sends the reply, and never receives it, what happens? queries when the route goes away, sets a SI-retransmit timer to one half the configured active time (default 90 seconds) fter this time has passed, sends an SI Query to If a SI Reply is sent to the SI Query, the timer is reset and the peer relationship stays up. If no SI Reply is sent, the peer relationship will reset SI Query will be sent a total of 3 times, and if no reply is received, the peer relationship will be reset /24 C C /24 Gone; No FS Query SI Query /24 Gone; No FS ad Link, Reply and SI Reply Never Makes It 40

39 Query or Not to Query The ctive Process Likewise, sends an SI Query to C and sets its SI-retransmit timer. If C fails to reply to the SI Query, will declare C as Stuck in ctive and reset its relationship with C. If C sends a Reply to, will Reply to clearing the Query process Either event clears the Query from and s point of view. If anything gets reset, it s the peer adjacent to the problem router, helping to troubleshoot and identify problem routers easier /24 C C /24 Gone; No FS Query SI Query /24 Gone; No FS SI Query Reset Relationship! SI Reply /24 Gone Remove /24 41

40 Query or Not to Query The Stuck in ctive Process SI-queries are sent to a peer up to three times May attempt to get a reply from a peer for a total of six minutes If still no reply by the end of this process, consider the route stuck through this peer On the router that doesn t get a reply after three SI-queries Reinitializes peer(s) who didn t answer Goes active on all routes known through bounced peer(s) Re-advertises to bounced peer all routes that we were advertising 42

41 Query or Not to Query Query ounding Where Does the Query Stop? /24 Router loses its connection to /24 Router does not consider a FS Router sends a query Router examines its local tables, and finds: Its current path (successor) doesn t pass through It has a FS that doesn t pass through Router answers The query is bounded where there is local knowledge of another loop-free path C D E F Local Knowledge of an lternate Path, So Reply G 43

42 Filter Query or Not to Query Query ounding Router C is filtering /24 towards D Router loses its connection to /24 Router sends C a query /24 Router C has no FS for /24 Router C sends D a query Router D examines its local tables No information about , so send an infinity reply Query is bounded because D has no information about /24 No Knowledge of Route, So Reply C D D E E F F G G 44

43 Summary Query or Not to Query Query ounding Router E is summarizing towards F Router loses its connection to /24 Router sends E a query Router E has no FS for /24 Router E sends F a query Router F examines its local tables No information about /24, so send an infinity reply Query is bounded because F has no information about /24 C D /24 E F G No Knowledge of Route, So Reply 45

44 Query or Not to Query Query ounding Router G has no other peers Router loses its connection to /24 Router sends G a query /24 Router G examines its local tables No FS (Feasible Successor) No peers to query, so send an infinity reply C E E G D F G F No peers, So Reply 46

45 /24 Query or Not to Query Query ounding - Stubs If these spokes are remote sites, often they have two connections for redundancy, not so they can transit traffic between and should not use the spokes as a path to anything not directly on the spokes, so there s no reason to learn about, or query for, routes through these spokes Don t Use These Paths For Transit 48

46 /24 Query or Not to Query Query ounding - Stubs Marking the spokes as stubs allows them to signal and that they are not transit paths will not query the stub remotes, reducing the total number of queries in this example to one Marking the remotes as stubs also reduces the complexity of this topology; now believes it only has one path to /24, rather than five 49

47 Query or Not to Query Query ounding - Stubs is manually configured as a stub sends Hello packets with peerinfo bits corresponding to the stubbing options: PEER_LLOWS_CONNECTED PEER_LLOWS_STTICS PEER_LLOWS_SUMMRYS PEER_IS_STUED PEER_LLOWS_REDIST PEER_LLOWS_LEKING receives the Hello with the stubbing options and marks as a Stub will then suppress all Queries to STU Hello Dst= PEERINFO=_options Hello Dst= HU 50

48 Classic and Wide Metrics

49 Classic and Wide Metrics Internal Routing Information What is an Internal Route in EIGRP? ny route within EIGRP that originated inside of the EIGRP process asically, routes which are associated with an interface which EIGRP has a network statement covering What is an External Route in EIGRP? Routes redistributed into EIGRP from another routing protocols, static routes, or connected routes EIGRP LWYS prefers Internal routes over External 52

50 Classic and Wide Metrics Computing Metrics Where does EIGRP get the vector metrics? andwidth: default bandwidth value or interface level bandwidth config Delay: default interface value or interface level delay config Reliability: per interface computed reliability, Load: per interface computed load, Why not set the K values so the reliability and load are picked up? Interface level computed metrics are only picked up when a change in the bandwidth or delay causes EIGRP to reread them or when a route changes and we have to recalculate the metric. Effectively, this means these metrics (reliability and load) are not checked on an ongoing basis with stable routes. 54

51 Classic and Wide Metrics Computing Classic Metrics EIGRP s calculated metric is called the composite metric Its computed from individual metrics called vector metrics - minimum bandwidth, total delay, load, reliability Interface metrics are converted before use bandwidth (in kilobits per second): 10 7 / Interface bandwidth delay (in 10s of microseconds): interface delay / 10ms load, reliability: converted to range of metric = [(K 1 bandwidth + K 2 bandwidth + (K 3 Delay)) K Load K 4 + Reliability ] 256 Constants (K 1 through K 5 ) are used to control the computation Default K values are: K 1 == K 3 == 1 and K 2 == K 4 == K 5 == 0 When K 5 is equal to 0 then [K 5 /( K 4 + reliability)] is defined to be 1 55

52 Classic and Wide Metrics Computing Metrics Router advertises /24 to andwidth is set to 1000 Delay is set to 100 Router Compares current bandwidth to bandwidth of link to ; sets bandwidth to 100 dds delay along link to, for a total of 1100 Router C Compares current bandwidth to bandwidth of link to ; sets bandwidth to 56 dds delay along link to, for a total of /24 é ê ëmin W: 1000 Delay: 100 W: 100 Delay: 1100 Minimum W: 100 Delay: ù + ådelaysú * 256 û ( bandwidth) W: 56 Delay: 3100 dded Together W: 56 Delay: 2000 C 56

53 Classic and Wide Metrics Computing Classic Metrics Router C uses the formula to compute a composite metric - This isn t what the router computes, though why? - The router drops the remainder after the first step! Why the 256? EIGRP uses a 32-bit metric space IGRP used a 24-bit metric space To convert between the two, multiply or divide by 256! é ê ëmin 10 7 ù + ådelaysú * 256 û ( bandwidth) é 10 7 ù ê ú * = ë û æ10 7 ç è 56 = ö ø ( )* 256 = ??

54 Classic and Wide Metrics Computing Wide Metrics EIGRP still used vector metrics, but they are not scaled, and are processed differently [(K 1 Throughput + { New vector metrics are derived from values reported by router Throughput derived from interface bandwidth Latency derived from interface delay Load derived from interface load Reliability derived from interface reliability K 2 Throughput Extended Metrics derived from router and/or configuration }) + (K 3 Latency) + (K 6 Ext Metrics) ] Load K 4 + Reliability K 5 Constants (K 1 through K 6 ) are used to control the computation Default K values are: K 1 == K 3 == 1 and K 2 == K 4 == K 5 == K 6 == 0 58

55 Classic and Wide Metrics Computing Wide Metrics y default, EIGRP computes throughput using the maximum theoretical throughput The formula for the conversion for max-throughput value directly from the interface without consideration of congestion-based effects is as follows: Max-Throughput = (K 1 If K 2 is used, the effect of congestion, as a measure of load reported by the interface, will be used to simulate the available throughput, by adjusting the maximum throughput according to the formula: Net-Throughput = [Max-Throughput + ( EIGRP_NDWIDTH EIGRP_WIDE_SCLE andwidth K 2 Max-Throughput Load This inversion of bandwidth value results in a larger number (more time), ultimately generating a worse metric. The inverted value is used only by the local router, the original bandwidth value is send to its neighbors )] ) 59

56 Classic and Wide Metrics Computing Wide Metrics K 3 is used to allow latency-based path selection. Latency and delay are similar terms that refer to the amount of time it takes a bit to be transmitted to an adjacent peer. EIGRP uses one-way based latency values provided either by IOS interfaces or computed as a factor of the links bandwidth Latency = (K 3 Delay EIGRP_WIDE_SCLE EIGRP_DELY_PICO ) For IOS interfaces that do not exceed 1 gigabit, this value will be derived from the reported interface delay, converted to picoseconds Delay = ( Interface Delay EIGRP_DELY_PICO ) For IOS interfaces beyond 1 gigabit, IOS does not report delays properly, therefore a computed delay value will be used Delay = ( EIGRP_NDWIDTH EIGRP_DELY_PICO Interface andwidth ) 60

57 EIGRP Packets

58 EIGRP Packets asic Packet Encoding Opcodes UPDTE 1 REQUEST 2 QUERY 3 REPLY 4 HELLO 5 CK 8 SIQUERY 10 SIREPLY 11 Packet Header Section TLV Data Header Opcode Checksum Version Flags Sequence Number CK Number VRID utonomous System TLV Encoding (variable length) Flags Init Flag (0x01) - This bit is set in the initial update packet sent to a newly discovered peer. It requests the peer to download a full set of routes CR Flag (0x02) - This bit indicates that receivers should only accept the packet if they are in Conditionally Received mode RS (0x04) - The Restart flag is set in the Hello and the Init update packets during the NSF signaling period EOT (0x08) - The End-of-Table flag marks the end of the startup updates sent to a new peer 62

59 EIGRP Packets Generic TLV Encoding Classic Wide Metric PRMETER 0x0001 0x0001 UTHENTICTION 0x0002 0x0002 SEQUENCE 0x0003 0x0003 SOFTWRE VERSION 0x0004 0x0004 MULTICST SEQUENCE 0x0005 0x0005 PEER INFORMTION 0x0006 0x0006 PEER TERMINTION 0x0007 0x0007 PEER TID LIST --- 0x Type Vector Data (Variable) Length Protocol ID VERSION General 0x0000 Classic IPv4 0x0100 Classic ppletalk 0x0200 Classic IPX 0x0300 Classic IPv6 0x0400 Classic ased on FI 0x0600 Multi-Protocol Generic TLVs apply to all address and service families Length includes the Type and Length fields Vector data is variable length 64

60 EIGRP Packets Example: Parameter TLV Encoding The Hello packet may carry the Parameter TLV to indicate the default coefficients(kvalues) should not used for computing the composite metric Opcode Length K1 K2 K3 K4 K5 K6 Hold Time Opcode transmitted as 0x0001, Length transmitted as 0x000C Number bytes in the Vector of the TLV.. Currently transmitted as 4 K-values - The K-values associated with the EIGRP composite metric equation. The default values for weights are: Hold Time - The amount of time in seconds that a receiving router should consider the sending peer valid. Default K values are: K 1 == K 3 == 1 and K 2 == K 4 == K 5 == K 6 == 0 65

61 EIGRP Packets Classic TLV Encoding Metric Encoding Scaled Delay ccumulative delay along an unloaded path to the destination. Expressed in units of 10 µsec 256 Scaled andwidth - Path bandwidth measured in bits per second In units of 10,000,000/kilobits per second 256 Metric Section Scaled Delay Scaled andwidth MTU Reliability Load Internal Tag Hop Count Opaque MTU - The minimum maximum transmission unit size for the path to the destination Hop Count - The number of router traversals to the destination Reliability - The current error rate for the path. Measured as an error percentage. value of 255 indicates 100% reliability Load - The load utilization of the path to the destination. Measured as a percentage of load. value of 255 indicates 100% load Internal-Tag - tag assigned by the network administrator that is untouched by EIGRP 66

62 EIGRP Packets Classic TLV Encoding Internal EIGRP Routes Used For: Internal IPv4 prefixes - TLV (Type 0x0102) Internal IPv6 prefixes - TLV (Type 0x0402) Subnet Mask it Count Type Next Hop Forwarding ddress Metric Section Length ddress (variable length) ((it_count - 1) / 8) + 1 Next Hop Forwarding ddress Specific address to use for the destination s nexthop. If the value is 0, the source address of the sending router is used as the next-hop for the route Metric Section ccumulative metric for destinations contained in this TLV Destination Section - The protocol specific address being sent 67

63 EIGRP Packets Classic TLV Encoding External EIGRP Routes Used For: External IPv4 prefixes - TLV (Type 0x0103) External IPv6 prefixes - TLV (Type 0x0403) External Protocol: Flags: Protocols Value IGRP 1 EIGRP 2 Static 3 RIP 4 HELLO 5 OSPF 6 ISIS 7 EGP 8 GP 9 IDRP 10 Connected 11 Exterior Information Type Length Next Hop Forwarding ddress Router ID S Number dministrator Tag External Protocol Metric Reserved External Flags Field Protocol Metric Section Subnet Mask it Count ddress (variable length) ((it_count - 1) / 8) + 1 Candidate Default (it 1) - If set, this destination should be regarded as a candidate for the default route. 68

64 EIGRP Packets Multi-Protocol TLV Encoding Common Type Code for all Protocols ll Internal prefixes - TLV (Type 0x0602) ll External prefixes - TLV (Type 0x0603) ll attributes are organized independent of the transport, or destination address/service family Optional Metric ttributes: Extended Metrics Extended Community Tags Exterior Information ddress Family Identifier (FI) - defines the IN type and format for the destination data. Topology ID (TID) 16bit number used to identify a specific sub-topology the prefixes is associated with Router ID (RID) unique 32bit number that identifies the router sourcing the route Destination Section - The address, as defined by the FI, being sent Type FI RID Metric Section Nexthop ddress External Information (Conditional) Destination (variable length) Length TID 70

65 EIGRP Packets Multi-Protocol TLV Encoding Metric Encoding Multi-Protocol metrics, which support the new Wide Metrics transmits Latency and Throughput values un-scaled Metric Section Offset Priority Reliability Load MTU Hop Count Latency Throughput Reserved Opaque Flags Extended Metrics In addition to the Classic vector metrics, there are new fields not previously available: Priority: Priority of the prefix when transmitting a group of destination addresses to neighboring routers. priority of zero indicates no priority is set. Currently transmitted as 0 Extended Metrics When present, defines extendable per destination attributes. This field is not normally transmitted 71

66 EIGRP Packets Multi-Protocol TLV Encoding Extended Metrics Extended Metrics is a way to influence routing decisions though the use of non-traditional vectors Currently there are 6 Extended Metric opcodes defined Value Description Type 1 Scaled Metric informational 2 dministrator Tag Policy/Filtering 3 Extended Community Tag(s) Policy/Filtering 4 Jitter Metric Modifier 5 Quiescent Energy Metric Modifier 6 Energy Metric Modifier 7 dd Path Nexthop Selection Extended Metric vectors that modify the composite metric, are added controlled by the composite K 6 73

67 What s Next!

68 What s Next! Multi-Vendor Support for EIGRP Ixia Emulation Software Wire shark Packet nalyzer Source Forge Easy-EIGRP lternate vendors offering EIGRP Easy-EIGRP (Sourceforge) 76

69 Recommended Reading for SIN: ISN ISN: Open EIGRP: draft-savage-eigrp-00 77

70 Complete Your Online Session Evaluation Give us your feedback and you could win fabulous prizes. Winners announced daily. Receive 20 Cisco Daily Challenge points for each session evaluation you complete. Complete your session evaluation online now through either the mobile app or internet kiosk stations. Maximize your Cisco Live experience with your free Cisco Live 365 account. Download session PDFs, view sessions on-demand and participate in live activities throughout the year. Click the Enter Cisco Live 365 button in your Cisco Live portal to log in. 78

71