Routers

Routers are devices which make it possible to "choose" the path that datagrams will take to arrive at the destination.
They are machines with several network interface cards each one of which is linked to a different network. So, in the simplest configuration, the router only has to "look at" what network a computer is located on to send datagrams to it from the originator.

However, on the Internet the schema is much more complicated for the following reasons:

• The number of networks to which a router is connected is generally large
• The networks to which the router is linked can be linked to other networks that the router cannot see directly

So, routers work using routing tables and protocols, according to the following model:

• The router receives a frame from a machine connected to one of the networks it is attached to
• Datagrams are sent on the IP layer
• The router looks at the datagram's header
• If the destination IP address belongs to one of the networks to which one of the router interfaces is attached, the information must be sent at layer 4 after the IP header has been unencapsulated (removed)
• If the destination IP address is part of a different network, the router consults its routing table, a table which establishes the path to take for a given address.
• The router sends the datagram using the network interface card linked to the network on which the router decides to send the packet.

In the case of indirect delivery, the role of the router and in particular that of the routing table is very important. So, the operation of a router is determined by the way in which this routing table is created.

• If the routing table is entered manually by the administrator, it is a static routing (suitable for small networks)
• If the router builds its own routing tables using information that it receives (via the routing protocols), it is a dynamic routing

The routing table

The routing table is a table of connections between the target machine address and the node according to which the router must deliver the message. In reality it is enough that the message is delivered to the network that contains the machine, it is therefore not necessary to store the complete IP address of the machine: only the network identifier of the IP address (i.e. the network ID) needs to be stored.

The routing table is therefore a table which contains address pairs:

Using this table, the router knowing the address of the recipient encapsulated in the message, will be able to find out what interface to send the message on (this comes back to knowing which network interface card to use), and to which router, directly accessible on the network to which this card is connected, to send the datagram.
This mechanism consisting of only knowing the address of the next link leading to the destination is called next-hop routing.

However, it may be that the recipient belongs to a non referenced network in the routing table. In this case, the router uses a default router (also called the default gateway).

Here, in a simplified way is what a routing table could look like:

The message is therefore sent from router to router by successive hops, until the recipient belongs to a network directly connected to a router. This then sends the message directly to the target machine...

In the case of static routing, it is the administrator who updates the routing table.
In the case of dynamic routing a protocol called a routing protocol enables the automatic updating of the table so that it contains the optimal route at any time.

Routing protocols

The internet is a collection of connected networks. As a result, all routers do not work in the same way, this depends on the type of network upon which they are found.

In fact, there are different levels of routers which operate with different protocols:

• Node routers are the main routers because they link the different networks
• External routers allow a connection to autonomous networks between them. They work with a protocol called EGP (Exterior Gateway Protocol) which develops gradually by keeping the same name
• Internal routers allow routing of information inside an autonomous network. They exchange information using protocols called IGP (Interior Gateway Protocol), such as RIP and OSPF

The RIP protocol

RIP means Routing Information Protocol. It is a Vector Distance type protocol, i.e. each router communicates to the other routers the distance which separates them (the number of hops which separates them). So, when a router receives one of these messages it increments this distance by 1 and sends the message to directly accessible routers. In this way, the routers can then keep the optimal route of a message by storing the next router address in the routing table in such a way that the number of hops to reach a network is kept to a minimum. However this protocol only takes into account the distance between two machines in terms of hops and does not consider the state of the connection so as to select the best possible bandwidth.

The OSPF protocol

OSPF (Open Shortest Path First) is more effective than RIP and is therefore beginning to gradually replace it. It is a protocol route-link type protocol; this means that contrary to RIP, this protocol does not send the number of hops which separates them to the adjacent routers, but the state of the connection which separates them. In this way, each router is capable of sending a card of the state of the network and can as a result choose the most appropriate route for a given message at any time.

In addition, this protocol avoids intermediary routers having to increment the number of hops, which results in much less extensive information making it possible to have more useful bandwidth than with RIP.

Last update on Thursday October 16, 2008 02:43:22 PM.This document entitled « Routing over the Internet » from Kioskea (en.kioskea.net) is made available under the Creative Commons license. You can copy, modify copies of this page, under the conditions stipulated by the licence, as this note appears clearly.