CSC 458/2209 – Computer Networks

Lecture 5: Routing
Professor Peter Marbach
Department of Computer Science
University of Toronto
Packet Routing and Forwarding
Forwarding IP datagrams
Class-based vs. CIDR

Routing Techniques
Naïve: Flooding
Distance vector: Distributed Bellman Ford Algorithm
Link state: Dijkstra’s Shortest Path First-based
Algorithm

Routing is a very complex subject, and has many aspects.

Here,
CSC 458/CSC 2209 – Computer we will concentrate on
Networks the
University basics.
of Toronto – Fall 2019 2
The Problem
“A” “B”

RR22

RR11 RR44

How does R1 choose a

next-hop on the path RR33
towards host B?

CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 3

What is Routing?
A famous quotation from RFC 791
 “A name indicates what we seek.
An address indicates where it is.
A route indicates how we get there.”
-- Jon Postel

CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 4

Forwarding vs. Routing
Forwarding: data plane
Directing a data packet to an outgoing link
Individual router using a forwarding table
Routing: control plane
Computing paths the packets will follow
Routers talking amongst themselves
Individual router creating a forwarding table

CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 5

Why Does Routing Matter?
End-to-end performance
Quality of the path affects user performance
Propagation delay, throughput, and packet loss
Use of network resources
Balance of the traffic over the routers and links
Avoiding congestion by directing traffic to lightly-
loaded links
Transient disruptions during changes
Failures, maintenance, and load balancing
Limiting packet loss and delay during changes

CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 6

 Example Network
Objective: Determine the route from A to B that minimizes the
path cost.
Examples of link cost:
A Distance, data rate, price,
congestion/delay, …

1 1 4
R1 R2 R4 R6

2 2 3
2
R7 3
R5 2
R3 4 B
R8

CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 7

Example Network
In this simple case, solution is clear from inspection
A

1 1 4
R1 R2 R4 R6

2 2 3
2
R7 3
R5 2
R3 4 B
R8

CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 8

What about this Network...!?

Learn more at
http://www.lumeta.com

CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 9

Technique 1: Naïve Approach
Flood! -- Routers forward packets to all ports
except the ingress port.

R1

Advantages:
 Simple
 Every destination in the network is reachable.
Disadvantages:
 Some routers receive a packet multiple times.
 Packets can go round in loops forever.
 Inefficient.

CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 10

Lowest Cost Routes
Objective: Find the lowest cost route from each of
(R1, …, R7) to R8.

1 1 4
R1 R2 R4 R6

2 2 3
2
R7 3
R5 2
R3 4
R8

CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 11

A Spanning Tree
1 1 4
R1 R2 R4 R6
3
2 2
2
R7
R5 2 3

4
R3 R8

 The solution is a spanning tree with R8 as the root of the tree.

 Tree: There are no loops.
 Spanning: All nodes included.
 We’ll see two algorithms that build spanning trees automatically:
 The distributed Bellman-Ford algorithm
 Dijkstra’s shortest path first algorithm

CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 12

Technique 2: Distance Vector
Distributed Bellman-Ford Algorithm
Define distances at each node x
 dx(y) = cost of least-cost path from x to y
Update distances based on neighbors
 dx(y) = min {c(x,v) + dv(y)} over all neighbors v

v 2 y
3 1
1
u
x 4 z
2 1
5 t du(z) = min{c(u,v) + dv(z),
w 4 3
s c(u,w) + dw(z)}

CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 13

Distance Vector Algorithm
c(x,v) = cost for direct link from x to v
Node x maintains costs of direct links c(x,v)
Dx(y) = estimate of least cost from x to y
Node x maintains distance vector Dx = [Dx(y): y є N ]
Node x maintains its neighbors’ distance vectors
For each neighbor v, x maintains Dv = [Dv(y): y є N ]

Each node v periodically sends Dv to its neighbors

And neighbors update their own distance vectors
Dx(y) ← minv{c(x,v) + Dv(y)} for each node y ∊ N N
Over time, the distance vector Dx

CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 14

Distance Vector Algorithm
Iterative, asynchronous:
each local iteration caused
Each node:
by:
Local link cost change
wait for (change in local link cost
Distance vector update or message from neighbor)
message from neighbor
Distributed:
recompute estimates
Each node notifies
neighbors only when its
DV changes if DV to any destination has
changed, notify neighbors
Neighbors then notify
their neighbors if
necessary
CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 15
Distance Vector Example: Step 1
Optimum 1-hop paths
Table for A Table for B
Dst Cst Hop Dst Cst Hop
E 3 C
A 0 A A 4 A 1

B 4 B B 0 B F 1
2
C  – C  – 6
D  – D 3 D 1
3 D
E 2 E E  – A 4
F 6 F F 1 F B

Table for C Table for D Table for E Table for F

Dst Cst Hop Dst Cst Hop Dst Cst Hop Dst Cst Hop
A  – A  – A 2 A A 6 A
B  – B 3 B B  – B 1 B
C 0 C C 1 C C  – C 1 C
D 1 D D 0 D D  – D  –
E  – E  – E 0 E E 3 E
F
CSC 458/CSC 1 – Computer
2209 F F
Networks  – F 3 F F of Toronto
University 0 –FFall 2019 16
Distance Vector Example: Step 2
Optimum 2-hop paths
Table for A Table for B
Dst Cst Hop Dst Cst Hop
A 0 A A 4 A E 3 C
1
B 4 B B 0 B
F 1
C 7 F C 2 F 2
D 7 B D 3 D 6
1
E 2 E E 4 F 3 D
F 5 E F 1 F A 4
B

Table for C Table for D Table for E Table for F

Dst Cst Hop Dst Cst Hop Dst Cst Hop Dst Cst Hop
A 7 F A 7 B A 2 A A 5 B
B 2 F B 3 B B 4 F B 1 B
C 0 C C 1 C C 4 F C 1 C
D 1 D D 0 D D  – D 2 C
E 4 F E  – E 0 E E 3 E
F 1 F F 2 C F 3 F F 0 F
CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 17
Distance Vector Example: Step 3
Optimum 3-hop paths

Table for A Table for B

Dst Cst Hop Dst Cst Hop
A 0 A A 4 A E 3 C
1
B 4 B B 0 B
F 1
C 6 E C 2 F 2
D 7 B D 3 D 6
1
E 2 E E 4 F 3 D
F 5 E F 1 F A 4
B

Table for C Table for D Table for E Table for F

Dst Cst Hop Dst Cst Hop Dst Cst Hop Dst Cst Hop
A 6 F A 7 B A 2 A A 5 B
B 2 F B 3 B B 4 F B 1 B
C 0 C C 1 C C 4 F C 1 C
D 1 D D 0 D D 5 F D 2 C
E 4 F E 5 C E 0 E E 3 E
F 1 F F 2 C F 3 F F 0 F
CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 18
Bellman-Ford Algorithm
Questions:
How long can the algorithm take to run?
How do we know that the algorithm always
converges?
What happens when link costs change, or when
routers/links fail?

Topology changes make life hard for the Bellman-

Ford algorithm…

CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 19

A Problem with Bellman-Ford
Bad news travels slowly

1 1 1
R1 R2 R3 R4

Consider the calculation of distances to R4:

Time R1 R2 R3
0 3,R2 2,R3 1, R4 R3 R4 fails
1 3,R2 2,R3 3,R2
2 3,R2 4,R3 3,R2
3 5,R2 4,R3 5,R2
… …
“Counting to…
infinity” …

CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 20

Counting to Infinity Problem – Solutions
Set infinity = “some small integer” (e.g. 16). Stop
when count = 16.
Split Horizon: Because R2 received lowest cost path
from R3, it does not advertise cost to R3
Split-horizon with poison reverse: R2 advertises
infinity to R3
There are many problems with (and fixes for) the
Bellman-Ford algorithm.

CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 21

Technique 3: Link State
Dijkstra’s Shortest Path First Algorithm

Routers send out update messages whenever the state of an

incident link changes.
Called “Link State Updates”

Based on all link state updates received each router

calculates lowest cost path to all others, starting from itself.
Use Dijkstra’s single-source shortest path algorithm
Assume all updates are consistent

At each step of the algorithm, router adds the next shortest
(i.e. lowest-cost) path to the tree.

Finds spanning tree rooted at the router.

CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 22
Dijsktra’s Algorithm
1 Initialization:
2 S = {u}
3 for all nodes v
4 if v adjacent to u {
5 D(v) = c(u,v)
6 else D(v) = ∞
7
8 Loop
9 find w not in S with the smallest D(w)
10 add w to S
11 update D(v) for all v adjacent to w and not in S:
12 D(v) = min{D(v), D(w) + c(w,v)}
13 until all nodes in S

CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 23

Dijkstra’s Algorithm Example
Find Routes for the Red (Leftmost) Node

2 2
3 1 3 1
1 1
4 4
2 1 2 1
5 5
4 3 4 3

2 2
3 1 3 1
1 1
4 4
2 1 2 1
5 5
4 3 4 3
CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 24
Dijkstra’s Algorithm Example
2 2
3 1 3 1
1 1
4 4
2 1 2 1
5 5
4 3 4 3

2 2
3 1 3 1
1 1
4 4
2 1 2 1
5 5
4 3 4 3
CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 25
Shortest-Path Tree

Shortest-path tree from u Forwarding table at u

v 2 y
link
3 1
1
u
x 4 z
v (u,v)
2 1 w (u,w)
5 t x (u,w)
w 4 3
s y (u,v)
z (u,v)
s (u,w)
t (u,w)
CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 26
Reliable Flooding of LSP
The Link State Packet:
The ID of the router that created the LSP
List of directly connected neighbors, and cost
Sequence number
TTL

Reliable Flooding
Resend LSP over all links other than incident link, if the
sequence number is newer. Otherwise drop it.

Link State Detection:

Link layer failure
Loss of “hello” packets

CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 27

Comparison of LS and DV algorithms
Message complexity Robustness: what happens if
LS: with n nodes, E links, O(nE) router malfunctions?
messages sent LS:
DV: exchange between Node can advertise
neighbors only incorrect link cost
Convergence time varies Each node computes only
its own table
DV:
Speed of Convergence DV node can advertise
LS: O(n2) algorithm requires incorrect path cost
O(nE) messages Each node’s table used by
DV: convergence time others (error propagates)
varies
May be routing loops
Count-to-infinity problem
CSC 458/CSC 2209 – Computer Networks University of Toronto – Fall 2019 28