EIGRP Administrative Distance, EIGRP Route Summarization, EIGRP Stub

The default administrative distance - AD of EIGRP internal route is 90 which is marked by the "D" symbol. The AD of an external route or considered as an external route using route distribution is 170 which is marked by the "D EX" symbol. There is also an eigrp summary route with the AD of 5. All of this can be shown with the #show ip route eigrp command.

To change the AD of EIGRP use the command:
(config)#router eigrp
(config-router)#distance eigrp

Auto summarization is a default feature at EIGRP. The purpose of this feature is used to automatically summarize the routes that are advertised through classfull network boundaries (EIGRP and RIPv2). It is therefore almost always necessary to turn this feature off in order make the router available to advertise classless networks routes by using the no auto-summary command. It is important to configure this route on the router sending the route. Remember that when applying the summarization, the adjecency goes down for a moment before it goes up again.

There are several advantages when applying a route summarization. A crucial advantage is that the route tables are getting smaller, because several routes are summarized. Another advantage is that CPU processing will be lighter, as the routes in the table are smaller.

To configure auto summarization use the following command:
(config)#int
(config-if)#ip summary-address eigrp <>

There are several ways to summarize routes where each route needs to have a specific bandwidth:

1. If there are multiple routes with the same CIRs, then add the amount of the CIRs together and use that as the minimum bandwidth needed. To set this use the command:
(config-if)#bandwidth

2. If there are multiple routes with different CIRs, then configure the route using a point to point (sub-)interface mode and assign each VC with the required CIR bandwidth value. To set this use the command on each sub-interface:
(config)#int serial point-to-point
(config-subif)#ip address
(config-subif)#bandwidth

3. There is also a possibility to configure a multipoint sub-interfaces. This can be done by adding all the CIRs from the routers. To set this use the command on the main router:
(config)#int serial multipoint
(config-subif)#bandwidth

The EIGRP can also be configured as a stub. However the stub has to be set only in the hub router. The spoke router does not need to be configured for stub (although possible) because the spoke router does not need to keep the routing table. It simply forwards all the routers to the hub router. This feature can also combat the SIA (stuck in active) problem because in stub networks, spoke routers will not be queried for routes when the hub does not have a feasible successor. To configure stub use the command:
(config)#router eigrp 100
(config-router)#eigrp stub

-- 10 February 2011 --

Variance, Feasible Distance & Advertising Distance

Variance is used when we want to apply unequal cost load sharing. Remember that EIGRP can support load sharing up to 6 links. In order for this load sharing to apply, the result of variance value (X) multiplied by the FD of the successor must be bigger than the FD of the feasible successor. Nevertheless, make sure that the variance value is as small as possible, since big variance value can bring some problems. When the variance command is applied, load sharing will be applied, but there is no change in the FD of the successor nor the feasible successor. This can be verified by the command #show ip eigrp topology.

Applying variance is easy using the command:
(config)#router eigrp (eigrp AS number)
(config-router)#variance

On the EIGRP topology table there are 2 values which is the FD (feasible distance) and the AD (advertised distance) or also known as reported distance.
- The FD (first value) is the full metric from the router to the destination network.
- The AD (Second value) is the metric from the next hop of the router to the destination network. Therefore, in most times the AD is lower than the FD.

If the AD has a bigger value than the FD, then there is a possibility of an routing loop. This rule is checked before the successor and feasible successor is applied to the routing table. If this condition is found, the router will not let that route take place into the election of a route or feasible route. Another condition that must be met for the potential feasible successor is that the AD of the feasible successor must be smaller than the FD of the successor. In a condition that where there are no feasible successors, the route on the router will go to the active state and in the meanwhile send DUAL Query Packets.

-- 1 February 2011 --

EIGRP topology states & SIA

There are a few Codes in the EIGRP Topology Table. This table can be shown with the command #show ip eigrp topology. There are 3 important codes to understand which are Active (A), Passive (P) and replay Status (r).

P : This code shows that the route is a Passive route which means that the route is NOT being calculated by the DUAL algorithm (not in process of calculating). Therefore, a route in this state is usable.

A : This code shows that the route is in an Active state which means that the route is currently being calculated by the DUAL algorithm. While being in this process, the route is not available for use. The DUAL calculation is usually short, so this Active state is also short and soon changes to the Passive Route (P). However, there are certain conditions when the route stays in this condition which is the SIA (Stuck In Active) condition. SIA happens when the router loses its successor and does not have a feasible successor. The router will send queries to the neighboring routers for the feasible successor and if there is no answer for a very long time from them, the neighbor relationship will be reset and SIA state will occur.

r : This code means that the query has been sent, but there is no response from the neighbor.

There are 4 possibilities why a route goes to SIA:

1. The link is a unidirectional link. The main router can send a packet, but it is not able to receive a packet back.

2. The destination router resource is not available. This can happen when the router's CPU processing processor is very high.

3. The destination router memory is corrupted, so that it cannot reply the query.

4. The link is a low quality link so that the reply cannot be received by the original router.


-- 1 February 2011 --

EIGRP part 2

In order to create an adjacency with the neighbor, the eigrp discovering process needs to be done first. The process of discovering a neighbor is as below:

1. The originating router sends a Hello packet using multicast to 224.0.0.10
2. The neigboring router will detect the Hello packet and responses with an Update packet using unicast (NOT multicast!). In the same time it will also send an hello packet to the originating router.
3. The originating router receives the update packet and sends an unicast Ack packet to the neighbor router while sending an update packet from its current configuration to the neighbor router in unicast.
4. The neighbor router will send an ack back to the originating router to confirm the receival of the update packet from the originating router.

EIGRP does not require the neighbors to have the same hello and hold time (the same as dead time in ospf) values which is the opposite in ospf, although this difference can create an instability in the connection between the adjacent routers. However, metric configurations (k-weights) and the AS number of the adjacent routers MUST be the same or the connection between the adjacent routers will not happen. Another difference between OSPF and EIGRP is that OSPF has a dead time = 3 x hello time whereas EIGRP has a hold time = 4 x hello time.

The hello packets are by default sent every 5 seconds on a high speed link such as Etherned and FDDI, whereas in low speed links (less than T1 speed such as ATM)

To change the hello time use the command:
(config)#interface
(config-if)#ip hello eigrp

To change the metrics weights use the command:
(config)#router eigrp // 0 is not allowed
(config-router)#metric weights 0


To make a router in an EIGRP network do the following command:
(config)#router eigrp
(config-router)#no auto summary
(config-router)#network // the network can be the primary network or the secondary network (See below)

To show the neighbors use the #show ip eigrp neighbor command

To debug the packets do #debug eigrp packets

EIGRP is also possible when the secondary addresses are applied on the routers. A secondary address is when 2 ip addresses are used on a single physical interface. However, the secondary addresses must be configured properly and the secondary subnets must be the same between the neigboring routers. If the secondary address subnet is configured on a router and the same subnet is configured on the neighboring router as primary, eigrp will not work because the primary addresses of both routers are not in the same subnet. The adjecency will be formed, but on the show ip eigrp neighbor command the adjecency is formed between the primary networks, not the secondary.

To put a secondary address on an interface do the command:
(config)#interface
(config-if)#ip address secondary

EIGRP part 1

EIGRP stands for Enhanced Interior Gateway Routing Protocol with an AD = 90. The EIGRP is an advanced version of IGRP and is cisco proprietary. EIGRP uses a DUAL algorithm, and has a fast convergence, faster than OSPF. EIGRP supports VLSM because EIGRP update packets contain a prefix length for each network. This protocol sends multicasts packets to 224.0.0.10 for dynamic neigbor discovery and uses the IP protocol 88.

EIGRP activates by defaul equal-cost load sharing up to 4 paths, but can be configured to a maximum of 6 paths. The command is:
(config)#router eigrp
(config-router)#maximum-paths

The DUAL (Diffusing Update Algorithm) is an algorithm which calculates the successors (main routes) and the feasible successors (backup route). Therefore, should the successors fail, the feasible successor is instantly ready for use.

EIGRP has 3 tables which is the standard routing table, the EIGRP topology table and the neighbor table. The routing table contains primary routes (successors) only, while the EIGRP topology stores both the successors and the feasible successors. The neighbor table stores information about the EIGRP neigboring routers.

EIGRP's metric is calculated based on 5 k-weights which are in order bandwidth, delay, load, reliability and MTU. However, only bandwidth and delay is set active by default (set to 1) while all other is set inactive (set to 0). The bandwidth and delay can be configured with the command:

(config)#interface
(config-if)#bandwidth
(config-if)#delay

EIGRP has 5 different type of packets which are:

1. Hello packets that are used for neigbor discovery and maintain the connection with the neighbors. This data uses UDP packets and is therefore unreliable.

2. Acknowledgement packets that are used from the originating router to confirm the update received from the neigboring routers. This data uses also UDP packets and is therefore unreliable.

3. Query packets are sent when the router loses its successor route and does not have a feasible successor. This packet uses Reliable Transport Protocol (RTP) and is therefore reliable.

4. Reply packets that is used to response a query packet and can also mean that a new route is found to the destination router. This packet uses RTP and is therefore reliable.

5. Update packets that is sent and received to build a routing table or when a change in the network happens. If the latter happens, this packet will only send the changing routes instead of all the routes from the originating router, thus making EIGRP more resource effecient. This packet is send using unicast to the destination router and uses RTP protocol.

To see these number of packets sent or received use the #show ip eigrp traffic command.

-- 27 January 2011 --

Distance Vector routing, RIPv1, RIPv2, IGRP, EIGRP

Distance vector routing protocol is a routing protocol that allows the router route an arriving packet to its distance. Distance vector uses Bellman-Ford Algorithm where it uses 2 methods to send the packet:

1. It calculates its distance to the destination by hop counting
2. It uses its vector to direct the router where to send its packet to (the next router or hop) in order to reach the destination

The DV protocol is suitable for smaller and stable environment because a DV protocol takes a long time to converge which is a drawback. Also a drawback is that DV cannot distinguish speeds of the lines. Therefore, a directly connected E1 line or a directly connected 56 kbps line will be considered the same. Another disadvantange is that RIP and IGRP broadcasts its routing table every 30 seconds and 90 seconds whether there is a change in the network topology or not. This behavior will waste unnecessary bandwidth. However, DV has its advantages which is less overhead and less computation.

To shorten the convergence of the networks and possible looping occurrences, route poisoning is applied. By using route poisoning, the maximum hop counts is limited to 15. When the hop count is 16, the packet will be dropped. Also, when a router is down, it will set the destination to it as infinity (16 hops) and will 'poisen' the route to that router.

Protocols that uses distance vector routing is RIPv1, RIPv2, IGRP and EIGRP.

The characteristics of RIPv1 are:

- updates broadcasts every 30 sec to 255.255.255.255 despite there is a change it its topology or not.
- uses classfull protocol thus not supporting VLSM
- equal cost sharing is set as a default feature
- max hop counts = 15 hops (16 hops is considered as infinite)
- does not support clear-text or MD5 authentication
- route max = 25 routes. For more than 25 routes, another packet must be generated.

The characteristics of RIPv2 are:

- updates multicasts every 30 sec to 224.0.0.9 despite there is a change it its topology or not.
- uses classless protocol thus supports VLSM
- equal cost sharing is set as a default feature
- max hop counts = 15 hops (16 hops is considered as infinite)
- does support clear-text or MD5 authentication. MD5 authentication is recommended since the password is encrypted.
- route max = 25 routes. For more than 25 routes, another packet must be generated.

The characteristics of IGRP are:

- updates broadcasts every 90 sec to 255.255.255.255 despite there is a change it its topology or not.
- uses classfull protocol thus not supporting VLSM
- this protocol is Cisco proprietary protocol
- equal cost can be enabled with variance command

The characteristics of EIGRP are:

- updates multicasts to 224.0.0.10. A full routing table multicast will be sent when an adjecency is formed. When there is a change in the network, only the change of the topology will be updated.
- uses DUAL algorithm
- equal cost can be enabled with variance command


-- 22 December 2010 --

Routing Table Operation

Routers use their routing tables to determine the outgoing route of the incoming packets. The routing table operation goes as follow in the following order:

1. The route with the longest prefix (or the shortest subnet mask) will be prioritized first, despite of its routing protocol. So if there is a RIP route of 172.10.10.0/28 and an OSPF route of 172.10.10.0/29, it will choose the OSPF route when sending outgoing packets because 172.10.10.0/29 (6 hosts) has a longer prefix and thus more specific hosts than the 172.10.10.0/28 (14 hosts) route.

2. If the route has the same prefix length, it will look for its administrative distance (AD) of the connection to the neighboring routes. Administrative distances are distances that are measured based on the reliability of the connection and routing protocol of one router to the other. Routes with a lower AD are considered more reliable and will be chosen over the other routes with a higher AD number.

For example, directly connected routes are considered very reliable and has therefore an AD=0. RIP routes are considered simple and not so reliable and has therefore an AD=120. OSPF are considered more reliable than RIP and is given an AD=110.

3. If (1) and (2) are the same for multiple routes, the router will look for its metrics according to its table. The metric is the cost of a route to go from the source address to the destination address. Metrics are different from each other's routing protocol.

RIP metrics uses hop counts where the path with the lowest hops of routers is considered as the shortest path. OSPF metric uses cost which is the inversely proportional bandwidth of the current connection and gets prioritized. Lower cost means a faster interface with higher speed and gets prioritized. IGRP uses a composite metric, based on a the composite of bandwidth, delay, load, reliability and max. MTU. By default, IGRP chooses its route based on bandwidth and delay only. Lowest composite metric means better connection and gets prioritized.

4. If (1) to (3) has the same values, then equal-cost load sharing will be applied, where the load will be equally shared between the multiple connections.

In any of the 4 cases above, only the best route is shown on the show ip route command. The alternative route is there when the best route is down, but the alternative route configuration will not be shown in the show ip route command.

To show the IP route type the command:
RX#show ip route

It will show the routing configuration. A B[C/D] via E (optional), F, G
where:
A = type of connection (direct, RIP, IGRP etc)
B = the destination group IP address
C = administrative distance
D = metric
E = next hop inbound interface (not the final destination)
F = length of connection
G = interface connected to

for example:

R 172.10.0.0 [120/5] via 10.20.30.40, 0:02:34, serial0

means that the connection uses RIP routing protocol to the destination 172.10.0.0 with the administrative distance of RIP=120 and the metric of 5 hops. 10.20.30.40 is the inline interface of the next hop and the router is already connected for 2 minutes and 34 seconds. The connection is connected to the serial0 of the router.

-- 6 December 2010 --