Routing

We first need to know what is a route…

Come on we all know what is a route!!!
A path that leads us to our destination is known as a route.

Just think Mac needs to go to his school. If does not know

how will he find out the route?

He will ask somebody isn't it?

Lets say he has police men to ask at some crossings who know about
some of the places near them and other policemen standing at other
crossings. The next slide displays such a situation. The arrows indicate
the places and policemen, a policeman knows. Lets see how Mac finds
out the route to his school!!!
 There you I want to
are go to
This block
R2 school
R4 Go this
way

Where
I don’t know
is my any one
school other than
Geeeeee…! R4 I will
send him via
!!! R4
I am here!!!
R6

R2 = 2 stops
R3 = 3 stops
so I will This I want to
send him via way go to
R2
R3 school
R5
R1
So, we can see that Mac has ultimately
found out the route!!!
We can also see that a path has been
established from Mac’s starting place to his
school.
Now let us figure out the various
components involved in this model.
Mac – Analogous to Data Packet
Blocks – Analogous to Distinct Networks
Police Men – Analogous to routers.
Now we can explain

Router – Router is a device that connects

different networks and also finds the
minimum cost path between them.
The fact is that a router never finds out the
full path. Doing this is practically not
possible.
It just redirects the data packet to the next
router (also referred to as ‘next hop’) on the
shortest path. We will now try to learn how.
What is meant by “shortest path” ?
Well, the path from one point to another is logically
weighted according to the distance. The weight is called
“metric”.
Metric is calculated on the basis of a number of factors. For
instance –

 Number of “Hops” – Number of routers the data will have

to pass through.

 Time required – The total time delay of the path.

 Cost – The path taken may cost some units at every

router.
So how do we estimate the metric?

To estimate the metric and find out the

shortest path on that basis, some
standardized algorithms may be used.

Before this we must understand a

mathematical model representing a network.
Let’s first define a network
A network (also called a graph) can be defined by
two sets –
 A set of nodes (or vertices – machines) V.
 A set of connections (Edges or links) E.
An edge may be defined by a pair of vertices. Thus for
an edge to exist at least two vertices must exist in V.
If number of vertices is n then the maximum no. of
edges is nC2.
Thus the sets E and V may be
V2 defined as –
E1 E4
E2 V = {V1, V2, V3, V4}
V1 V3

E3 E5 E = { (v1,v2), (v1,v3), (v1,v4),

V4
(v2,v3), (v3,v4) }
Other definitions
Directed Graph (or Digraph) – When the elements of E
are ordered pairs then we call the graph a directed graph.
This means, that while traversing through the edge, we can
take only a fixed direction, as implied by arrows.

Thus the sets E and V may be

V2 E4
E1 defined as –
E2
V1 V3 V = {V1, V2, V3, V4}
E3 V4 E5
E = { (v1,v2), (v3, v1), (v1,v4),
(v2,v3), (v4, v3) }
Weighted Graph – This is a kind of graph where the
element of E are triples with two vertices and a number
representing the weight (cost) of the edge.

Thus the sets E and V may be

V2 E4, 4
E1, 2 defined as –

V1
E2, 5
V3 V = {V1, V2, V3, V4}
E = { (v1,v2, 2), (v3, v1, 5),
E3, 1
V4 E5, 2
(v1,v4, 1), (v2,v3, 4),
(v4, v3, 2) }
Path – This is a part of the graph connecting two distinct vertices and has no
loops.
More fundoo – It consists of subsets of V and E where each element of V is
associated with some E.
V2 E4 The highlighted part is a path from
E1
V2 to V4
E2
V1 V3

E3 V4 E5
Tree – This is a graph or sub-graph where there exists maximum one path
between two nodes. This simply means that there is no loop in the graph.

The highlighted portion is a tree

V2 E4, 4
E1, 2

E2, 5
V1 V3
E3, 1
V4 E5, 2

Spanning Tree – A tree that covers all the vertices of the graph or network is
called a spanning tree.

V2 E4 The highlighted part is a spanning

E1
tree
E2
V1 V3

E3 V4 E5
Algorithms to find the shortest path
between two nodes.
Dijkstra’s Algorithm
The algorithm works as follows:
1. Label the start vertex's final value as 0 (as it is the origin), and label it
1.
2. Update the working values (if they have no working value yet, or if
the new working value is lower) of all the vertices that can be
reached directly from the last vertex labelled.
3. Choose the unlabelled vertex with the smallest working value, and
record its working value as its final value and record its order of
labellling.
4. Repeat steps 2 and 3 until the destination vertex is labelled.
The working value for the destination vertex is its best working value.
The shortest route can now be found by tracking back:
If vertex a lies on the route, then vertex b is the previous vertex if the
label at a - label at b =weight of edge ab
Example of using Dijkstra's algorithm
to get from A to C:

This is the graph given

Step 1 – Label the start vertex's final value as 0 (as it is the origin), and label it 1.
Step 2 – Update the working values of B,E and D.

Step 3 – Choose the vertex with the lowest working value (B), and record its
working value as its final value and the order in which it was labelled
 Step 4 Update all working values

5 Label the smallest working value vertex with no label (E),final value=working value.
Step 6 – Update working values (no change)
Step 7 – Choose smallest working value (D), label.

Step 8 – Update working values (No change)

Step 9 – Choose and label smallest working value, unlabelled, vertex. It is C, so C's
final working value is

To find the path:

7-5=2 and 7-1=6 so the path can either be ADC or ABC (it doesn't matter which
we choose).
There are a number of similar algorithms for finding the shortest path.
There is another called Floyd’s Algorithm in which we use a matrix
formation to find out the shortest paths, but right now let us concentrate
on basics.

So how does the router take advantage of these algorithms?

The routers actually communicate with their neighbours to create a
table called routing table

This is how a routing table looks like.

Destination Gateway Flags Ref Use Interface

127.0.0.1 127.0.0.1 UH 1 298 lo0
default 172.16.12.1 UG 2 50360
172.16.12.0 172.16.12.2 U 40 111379 le0
172.16.2.0 172.16.12.3 UG 4 1179
172.16.1.0 172.16.12.3 UG 10 1113
172.16.3.0 172.16.12.3 UG 2 1379
172.16.4.0 172.16.12.3 UG 4 1119
If we look at the above table closely, we will find that it has the next hop
information. For instance if some packet is destined for the network
172.16.12.0 then it will be sent to the gateway 172.16.12.2 . The rest of
the headache is to be taken by the later.

A question may be asked over here that “how was this routing table
generated?”.
The routers have communicated the costs of connecting to their
neighbours to all the other routers. Then the routers have used some
or other “shortest path algorithms” to determine the routing table for
itself. This communication may be done only once (The first time they
are connected – Static Routing), Periodically (At fixed time intervals -
Dynamic) or in Triggered manner (Whenever there is some change in
the network – new router is added or an existing one is removed etc. -
Triggered).

Table updating can also be done manually in some routers. However in

a vast network like the internet, we would like automatic configurations.
Are there any standard methods for
communication between routers?
Yes, of course there are!!!
The simplest method is Flooding –
A router sends a copy of it’s table to its neighbours.
These neighbours update the table accordingly and send it
to their neighbours and this way, all the routers in the
network share their data. However this leads in congestion
in the network and leads to multiple copies of the same
data reaching on place.

To avoid this problem, occurring on a periodic basis,

triggered updating schemes are used.

Moreover time stamps are attached with these packets so

that old or already received packets can be rejected.
There are many protocols that define rules
for the communication between routers.
E.g. RIP, BGP etc.

Two important definitions -

5. Routing Protocols – The protocols that
govern how the routing should be
managed
6. Routed Protocols – The protocols that
govern the data flow.