Scribd
Upload
99 views

3-5 Simple Queue Model

This document discusses key concepts in computer networks including queue modeling, packet switching, and statistical multiplexing. It introduces a simple deterministic queue model to analyze packet behavior. Packet switching is explained as breaking messages into packets to allow parallel transmission across links, reducing end-to-end latency. Statistical multiplexing allows efficient sharing of a single link between multiple flows by taking advantage of periods when aggregate demand is below the link capacity.

Uploaded by

Brian
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
99 views

3-5 Simple Queue Model

This document discusses key concepts in computer networks including queue modeling, packet switching, and statistical multiplexing. It introduces a simple deterministic queue model to analyze packet behavior. Packet switching is explained as breaking messages into packets to allow parallel transmission across links, reducing end-to-end latency. Statistical multiplexing allows efficient sharing of a single link between multiple flows by taking advantage of periods when aggregate demand is below the link capacity.

Uploaded by

Brian
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 15

CS144

An Introduction to Computer
Networks
Packet Switching
Queue models

Nick McKeown
Professor of Electrical Engineering
and Computer Science, Stanford University

CS144, Stanford
University

Outline
1. Simple deterministic queue model
2. Small packets reduce end to end delay
3. Statistical multiplexing

CS144, Stanford
University

Simple model of a router


queue

CS144, Stanford
University

Simple model of a
queue
Cumulative number of
bytes arrived up until
A(t) t.
time

Q(t)
R Link
rate
D(t)

Cumulative number of
bytes departed up until
time t.
CS144, Stanford
University

Simple model of a
queue

CS144, Stanford
University

Every second, a 100 bit packet


arrives to a queue at rate
1000b/s. The maximum
departure rate is 500b/s. What
is the average occupancy of the
queue?

Cumulative
number of bits

Example
A(t)

100

D(t)

0.1s

0.2s

1.0s

Solution: During each repeating 1s cycle, the queue


fills at rate 500b/s for 0.1s, then drains at rate 500b/s
for 0.1s. Over the first 0.2s, the average queue
occupancy0.5
is therefore
bits.
(0.1 500) 25
The queue is empty for 0.8s every cycle, and so
average queue occupancy:
Q(t) (0.2 25) (0.8 0) 5
CS144, Stanford
University

time

Outline
1. Simple deterministic queue model
2. Small packets reduce end to end delay
3. Statistical multiplexing

CS144, Stanford
University

10

Packet Switching
Why not send the entire message in one packet?
A

p/r1

M/r1

R1

R1

R2

R2

R3

R3

M li
End-to-end delay, t

c
i ri

p li M p
End-to-end delay, t
1
c p rmin
i ri

Breaking message into packets allows parallel transmission


across all links, reducing end to end latency.
CS144, Stanford
University

11

Outline
1. Simple deterministic queue model
2. Small packets reduce end to end delay
3. Statistical multiplexing

CS144, Stanford
University

12

Statistical Multiplexing
Basic idea

Two flows

rate

Average
rate

One flow
rate

time
rate

time

CS144, Stanford
University

Many flows

Average rates of:


1, 2, 10, 100, 1000
flows.

time

Packet Switching
Statistical Multiplexing
Packets for
one output
1

Data

Hdr

Data

Hdr

R
R

Q(t)
Link rate, R

R
N

Data

Hdr

Packet
buffer

Statistical multiplexing means the egress link need not run


at rate NR.
The buffer absorbs brief periods when the aggregate rate
exceeds R.
Because the buffer has finite size losses can occur.

CS144, Stanford
University

15

Statistical Multiplexing
Rate

C
A

time

Rate

time
CS144, Stanford
University

16

Rate

Statistical Multiplexing
Gain
A+B
2C
R < 2C

B
time

Statistical multiplexing gain = 2C/R

CS144, Stanford
University

17

Summary
Often, we can use a simple deterministic
model of a queue to understand the
packet dynamics in a network.
We break messages into packets because
it lets us pipeline the transfer, and reduce
end to end delay.
Statistical multiplexing lets us carry many
flows efficiently on a single link.
CS144, Stanford
University

18

<end>

CS144, Stanford
University

19

You might also like