EIGRP Basics

Here’s a quick run down of some EIGRP basics every Network Engineer should know.

EIGRP was originally a Cisco proprietary protocol, however in 2013 EIGRP was published as an IETF Internet Draft called draft-savage-eigrp In May 2016 if became RFC 7868

EIGRP uses protocol number 88. It does not use UDP or TCP, but uses it’s own Reliable Transport Protocol (RTP). RTP allows routers to initially exchange full routing information, and then afterwards only routing changes.

RTP normally works over multicast, but can also work over unicast. RTP makes sure all routes are delivered reliably. The “Hello Interval” is the interval hellos are sent. The Hello protocol is used to maintain adjacencies, and does not contain any routing information.

The “Hold Timer” is the period of time a router waits to receive a hello, before determining the neighbour is dead.

Full updates are sent when a neighbour is discovered, otherwise partial updates are sent. 224.0.0.10 and FF02::A are the update multicast addresses.

• EIGRP supports all of the following:
• MD5 and SHA based authentication.
• Subnet mask in updates
• Route tags
• Advertisement of routes with a different next hop router than then advertising router.
• Route summarisation at any point in the network
• IIPv4 and IPv6

To maintain a loop free network EIGRP uses a so called Feasibility Condition to identify neighbours. Should a EIGRP router detect a topology change, it will query its neighbour using a “diffusing algorithm” to determine the new best paths. Neighbours who are also affected will propagate the query further, hence the search for a new path to an affected part of a network diffuses through the network.

To handle multiple topology changes EIGRP uses DUAL (Diffusing Update Algorithm) whih controls the diffusing algorithms, and inserts the appropriate information into the routing table.

EIGRP has a max hop count of 100 by default, but this can be adjusted with the “metric maximum-hops” command.

EIGRP uses internal and external routes. Internal routes are injected via the network command. External routes are redistributed. Internal routes have an AD of 90, External routes have an AD of 170. Administrative distances can be changed with the “distance eigrp” command.

EIGRP uses the following “classic metrics:

Bandwidth, Delay, Reliability, Load, MTU, and Hop Count.

The first 4 are bound together to make a single number called a “composite metric”. Newer versions of IOS support so-called “wide metrics” which expand on the classic metrics. The metrics are further explored below:

Bandwidth Metric Component

The bandwidth metric is assigned to each interface via the bandwidth command (or its default value). When calculating the composite metric, EIGRP uses the minimal bandwidth along the route.  For two connected interfaces the lower bandwidth is used. With classic metrics, EIGRP uses a range of 1Kbps to 10Gbps

Delay Metric Component

The delay metric is assigned to each interface via the delay command (or its default value).

The delay command uses tens of microseconds, so the command “delay 10” will be 100 microseconds. The show interface will show the delay in tens of microseconds, so in this case it will be show as 100. When calculating the composite metric, EIGRP uses the total delay value. For two connected interfaces, the sum of the delays are used. With classic metrics, EIGRP uses a range of 10  microseconds to 167,772,140 microseconds. A delay of 167,777,215 tens of microseconds is used to indicate an infinite distance and is used to indicate an unreachable route.

Reliability Metric Component

The reliability metric can be seen on the interface and is a value out of 255, where 255 is 100% reliable. The minimal reliability between two connected interfaces is used. EIGRP does not send updates when an interfaces reliability changes.

Load Metric

The load metric can be seen on an interface as Txload and Rxload and is a value out of 255, where 1 is no load and 255 is full load. EIGRP uses the maximum load between two connected interfaces. EIGRP does not send updates when an interfaces load changes.

MTU Metric

EIGRP advertises the minimum MTU along the route to the destination, but the use of MTU in the best path process has never been implemented.

Hop Count Metric

The hop count is simply the number of routers to a destination. Hop count is not used in the best path calculation in any way. Hop count is a safety measure used to avoid routing loops, by setting the maximum hop count. The default is 100, but it can be set from 1 to 255.

The following diagram summarises the uses of the metrics