Vanessa Avalos
Cisco Networking Academy Semester 3
Chapter 5
Routing Protocols: IGRP

The network layer provides best effort, end-to-end packet delivery services to its user, transport layer

Routing is the process of choosing the best path and how to cross multiple physical networks.  A router’s path determination function evaluates available paths to destinations and to establish the best path for routing a packet

When a router receives a packet:
   The packet is forwarded based on ip address
   After the path decision is made the router switches the packet out the correct port

How a routing table works:
   The routing table stores information on the destination and how to reach it
   Only the network portion of the ip address is stored, routing tables also store other information, depending on routing protocol.
   Routers have to communicate with each other in order to build routing tables.  They communicate by sending updates.  These updates can be all or part of a routing table.

Each routing protocol interprets the best path in it’s own way, but they all use metrics.
Usually, the smaller the metric, the better the path.

Some common metrics are:

Hop count Number of router a packet must go through to reach destination
Bandwidth Data capacity of link
Delay Time it takes to move a packet from source to destination
Load  Amount of activity on network resources such as router or link
Reach ability Error rates of each link
Ticks Delay on link using clock ticks
Cost Arbitrary value placed by network administrator

Addressing:
Addresses have a network part and a host part.
The network part is used in routing.  The router checks the packet destination to find out if it know what to do with it.  If the router knows what to do with it, it will change the physical address and send it on it’s way.  If there is no default route and the router doesn’t know where the packet goes, the packet is dropped.

Routed vs Routing:
Routed, done in routing, can be moved over a network.

Routed: IP & IPX
Routing: IGRP, EIGRP, RIP, BGP, OSPF, IS-IS

Multiple protocol routing:
2 classes of routing protocols
Interior: IGRP, EIGRP, OSPF, RIP
Interior protocols must have a list of associated networks before routing can start.
Exterior: BGP, IS-IS
Exterior protocols need a list of neighbors with which to exchange routing table information with and a list of networks to advertise as directly reachable.

The goals of a routing protocol:
Optimal Route
Simplicity and Efficiency
Roubustness
Rapid Convergence
Flexibility
Dynamic

Classes of interior routing protocols:
Distance Vector – like RIP and IGRP
Link state – like OSPF, IS-IS and NLSP
Hybrid  - like EIGRP

IGRP
Configuring IGRP
1) create the routing process with the router commands
2) Configure the protocol specifics
3) Assign an autonomous number (AS)

When choosing a routing protocol:
Make sure you consider your network size and complexity, traffic levels, security needs, reliability needs, organizational policies and organizations adaptability to change.

Distance Vector routing:
Distance Vector protocols send all or a portion of their routing tables to their neighbors.
IGRP a distance vector protocol uses the metrics: Delay, Bandwidth, Reliability and Load.  The defaults are Delay and Bandwidth.
Reliability and load can have a value from 1 to 25
Bandwidth can have values reflecting speed 1200bps to 10Gbps
Delay can have a value from 1 to 244
Using these options a network administrator can influence route selection.

Autonomous Systems
A group of routers under one admin/group

IGRP advertises Interior routes (routes between subnets in the network attached to a router interface, If not subnetted IGRP does not advertise); System routes (routes to other major networks with the same AS); and Exterior routes (are routes outside the AS: identified by gateway of last resort).

IGRP has a number of features designed to enhance stability:

Holddowns tell routers to hold any changes for a period of time usually just greater then the period of time necessary to update entire network.

The route to that network is placed into holddown when a router learns that a network is further away than it was previously know or it learns that a network is down.
During holddown time, the route is advertised but router does not accept updates about the route, except the router that originally advertised the new metric.
Holddown is a way of avoiding routing loops, but it can increase convergence time
They are used to prevent regular update messages from reinstating a route that may have gone bad. 

Split-horizon helps prevent routing loops between adjacent routers.

Split horizons occur when a router tries to send information about a route back to the router it came from.

Poison Reverse Updates are intended to prevent larger routing loops.
An increase in metrics generally indicates routing loops
Poison reverse updates are then sent to remove the route and place it in holddown
A router poisons a route by sending an update with a metric of infinity to a router that originally advertised route.  This helps speed convergence