Documente Academic
Documente Profesional
Documente Cultură
1-800-COURSES
www.globalknowledge.com
Introduction
All advanced IP routing protocols (OSPF, EIGRP, IS-IS) have several elements that are always present. This is
because the function of all routing protocols is the same, which is to find the best path to an IP destination
address.
The four common elements of Enhanced Interior Gateway Routing Protocol (EIGRP) are:
1. The operations, processes, and rules for the exchange of packets to accomplish the routing protocol
objective.
2. The construction and maintenance of the routing protocol database.
3. The construction and use of the data structures created by elements one and two.
4. The configuration and verification of the entire process.
The purpose of this paper is to provide an overview of the elements for the implementation of EIGRP.
IP
Header
Protocol
Number
Packet Payload
C
R
C
EIGRP was developed by Cisco as a proprietary routing protocol for Cisco devices. Cisco has recently made EIGRP
open. Everything for a routing protocol starts with the initial exchange of packets. The packets are transported in
an IP datagram using protocol number 88.
EIGRP Packets
Hello: Establish neighbor relationships
Update: Send routing updates
Query: Ask neighbors about routing information
Query Reply: Respond to query about routing
information
ACK: Acknowledge a reliable packet
The graphic shows the EIGRP packet types which include hello, update, query, query reply, and
acknowledgement. The use of the information carried in these packets will be described later. The exchange of
packets facilitates the following:
From the database, additional data structures are createdNeighbor Table, Interface Table, Routing Information
Base (RIB), Forwarding Information Base (FIB), and Adjacency database.
EIGRP is a reliable protocol because the query, query reply, and update packets are acknowledged and
retransmitted if necessary when errors are detected.
The graphic shows an initial exchange of hello packets to form a neighbor relationship between two routers
sometimes called an adjacency. For the neighbor relationship to be formed, several parameters in the hello
packet must match at both ends. The default interval for the EIGRP hello is five seconds with a dead interval of
15 seconds. If 15 seconds go by without a hello, the neighbor relationship is torn down.
From this beginning, the remaining operations will take place one by one.
The Cisco DUAL algorithm defines all the rules and processes that make this work, which include:
Pacing parameters for how many updates can be sent one after the other
1.
Neighbor index
2.
Neighbor IP address
3.
4.
5.
Neighbor uptime
6.
7.
Retransmission timeout
8.
9.
RTO
Q Seq
Cnt Num
2280 0 5
All of these parameters can be verified with the show ip eigrp interface and show ip eigrp neighbor commands.
EIGRP Metric
The use of metric components is represented by K
values
Metric components are:
Bandwidth (K1)
Delay (K3)
Reliability (K4 and K5)
Loading (K2)
The primary purpose of any IP routing protocol is to provide the best path to an IP destination address (best
is expressed as a number called a metric).
The metric is calculated in a different way by each routing protocol. With EIGRP, the DUAL algorithm defines the
EIGRP composite metric as shown in the graphic. The variables in the EIGRP metric calculation are shown in the
following list:
DUAL calculations use the bandwidth and delay variables in an equation to determine a best metric for each
destination. This information is sent to each neighbor as part of the routing update.
Router 5 is making the routing domain aware of network 10.1.1.0/24. Distance is expressed as the composite
metric described earlier. Network 10.1.1.0 is advertised by R5 to Routers 1 and 4 with an advertised distance
(metric) of 500. The link from R5 to both R1 and R4 has a metric of 500. Therefore, the feasible distance to
Copyright 2014 Global Knowledge Training LLC. All rights reserved.
10.1.1.0 from R1 and R4 is AD + Link Distance, or 500 + 500 = 1000. This FD of R1 and R4 now becomes an AD
(advertised distance) as they send it to other neighbors.
In the drawing, AD1 is the advertisement from R1 to R4. R4 will then advertise the route to R3 with an AD
(advertised distance) of 1000. Since the link from R1 to R3 is Ethernet, the link cost is 1000, so the feasible
distance (FD1) for R3 via R1 is 2000 (AD + LD, 1000 + 1000).
The route as advertised by R5 to R4 is AD2, and when the same process is used for the calculations at R3, FD2 will
be 500 + 500 + 500 + 1000 = 2500. In routing, the lowest metric is always preferred, and FD1 has a lower metric
(2000 vs. 2500).
These results are shown in the graphic and are used to construct the EIGRP topology table, which is the
compilation of all the routes received from ALL of the EIGRP neighbors and their associated advertised and
feasible distances. The routing table will be described later.
An example of an EIGRP topology table is shown in the graphic. Note that the P on the left means passive or
stable. Anything other than a P indicates that some sort of convergence event is in progress. Also note that the
best route is called a successor.
If multiple successors exist for the same network and all have an identical FD metric, load balancing (or Equal
Cost Multi-Pathing [ECMP]) will occur. The calculated successor will be copied to the Routing Table (or Routing
Information Base [RIB]).
The outbound interface is shown, and in this case it is serial 0/0/0.4.
The fraction 2297856/128256 is the FD/AD for the destination.
Since EIGRP is an advance distance vector routing protocol rather than a link-state routing protocol, there is the
possibility of a routing malfunction called a routing loop. Routing loops are usually caused by the attempt to
select an alternate route that includes a failed component being advertised by a neighbor in error. Over the
years, numerous rules have been devised to solve this problem.
EIGRP implements the split horizon rule that prevents networks from being advertised out of the same interface
where they were learned in the first place.
The DUAL algorithm in EIGRP implements another rule called the feasible condition.
Since there is a topology database recording the best routes, the next best routes, the next after that, (and so
forth), the feasible condition is applied to make sure that none of the alternative routes could cause a loop. The
ability to change to a second best route immediately upon the failure of a route is a very good feature for fast
convergence, but it must always be loop-free.
The feasible condition is that the advertised distance (AD) of an alternative route must be less than the current
feasible distance (FD) being used to reach the route.
Copyright 2014 Global Knowledge Training LLC. All rights reserved.
When this condition is met, the alternate route is called a feasible successor (FS).
If an alternate route exists and does not meet the feasible condition, then it may still be selected and installed as
a route after an EIGRP query process in order to insure that it is loop-free.
DUAL Operation
The topology table is changed when:
The cost or state of a directly connected link changes
An EIGRP packet (update, query, reply) is received
A neighbor is lost
The last bullet in the graphic indicates an ACTIVE condition. The EIGRP query process is part of the ACTIVE
condition.
A query packet is sent to each neighbor asking about the lost route. Each query is acknowledged and a reply
must be received for all queries, otherwise, convergence is not complete.
Copyright 2014 Global Knowledge Training LLC. All rights reserved.
A runaway query process is one of the things that can slow down EIGRP. Many rules and features scope (limit)
this process, such as proper route summarization and the EIGRP stub router feature (neither will be described in
detail).
The graphic provides a review of all the EIGRP processes described so far.
10
From the topology, all the successors (best routes) are copied to the Routing Information Base (RIB), which is also
called the routing table. This dynamic table is maintained as part of the routing process memory by incremental
routing updates after the initial loading process. The exchange of information necessary to maintain this table
is called the control plane.
Once the tables are converged and stable, it is necessary to quickly deal with changes in network topology (such
as link failures and device failures). With all the real-time protocols in modern networks (voice, video, multicast),
repair times are measured in milliseconds. Over the years, Cisco has evolved multiple features in the Cisco IOS to
address this issue.
11
Exchange of
Routing Information
CEF
Control Plane
Routing Protocol
IP Routing Table (RIB)
Data Plane
IP Forwarding Table (FIB)
Stored in fast
memory cache or
hardware
The latest feature in this evolution is Cisco Express Forwarding (CEF), which copies all the contents of the RIB
(routing table) to a fast cache in the fastest memory location possible, such as a line module in a 6500 or Nexus
multi-layer switch. This table is called the Forwarding Information Base (FIB) and it is located in the data plane.
This table is consulted by the router when packets arrive on an interface and a routing decision is required. The
FIB is refreshed by the RIB regularly. The only time the RIB is consulted is when a FIB fails to provide a routing
entry.
12
13
RTO
Q Seq
Cnt Num
2280 0 5
As described earlier, the show ip eigrp neighbor command is a logical way to verify that the EIGRP router is
seeing its neighbors.
The show ip route output will verify that EIGRP control plane information is being received from neighbors.
Debug commands may provide more detail on this process, but they must be used with caution.
14
Show ip protocols is perhaps the single most productive show command for verifying the entire EIGRP process
because it can show:
15
Mean
SRTT
0
10
Pacing Time
Un/Reliable
0/10
10/380
Multicast
Flow Timer
0
424
Pending
Routes
0
0
The show ip eigrp interface and show ip eigrp topology were described earlier and are important verification
tools as well.
16
Conclusion
EIGRP has been used successfully for many years by small, medium, and large enterprises that use mostly Cisco
devices. It converges as fast as any other internal gateway protocol and provides features that facilitate design,
implementation, and troubleshooting. Since it is proprietary, excellent support is provided by Cisco. So, this is a
viable choice for an enterprise routing protocol along with OSPF and IS-IS.
Learn More
Learn more about how you can improve productivity, enhance efficiency, and sharpen your competitive edge
through training.
ROUTE - Implementing Cisco IP Routing v1.0
ARCH - Designing Cisco Network Service Architectures v2.1
SPNGN1 - Building Cisco Service Provider Next-Generation Networks, Part 1
SPNGN2 - Building Cisco Service Provider Next-Generation Networks, Part 2
Visit www.globalknowledge.com or call 1-800-COURSES (1-800-268-7737) to speak with a Global Knowledge
training advisor.
17