Congestion Control and Quality of Service

23.1 Data Traffic

23.2 Congestion
23.3 Congestion Control
23.4 Two Examples
23.5 Quality of Service
23.6 Techniques to Improve QoS
23.7 Integrated Services
23.8 Differentiated Services
23.9 QoS in Switched Networks

Computer Networks 24-1

 DATA TRAFFIC
• Congestion control and quality of service are two issues so closely
bound together that improving one means improving the other and
ignoring one usually means ignoring the other.

• The main focus of congestion control and quality of service is data

traffic.
• In congestion control we try to avoid traffic congestion.
• In quality of service, we try to create an appropriate environment
for the traffic.

Computer Networks 24-2

 DATA TRAFFIC
Traffic Descriptors
• Traffic descriptor are qualitative values that represent a data flow
• Average data rate = amount of data/time
• Peak data rate: the max. data rate of the traffic
• Max. burst size: the max. length of time the traffic is generated at the peak rate
• Effective bandwidth: bandwidth that the network needs to allocate for traffic flow

Computer Networks 24-3

 Traffic Profiles
• Constant-bit-rate (CBR)
• Variable-bit-rate (VBR)
• Bursty

Computer Networks 24-4

 CONGESTION
• Congestion in a network may occur if the load on the network - the number of
packets sent to the network-is greater than the capacity of the network - the number
of packets a network can handle.
• Congestion control: the mechanisms to control the congestion and keep the load below
the capacity.
• Congestion occurs because routers and switches have queues- buffers that hold the
packets before and after processing.
• The rate of packet arrival > packet processing time  input queue longer
• The packet departure time < packet processing time  output queue longer

QUEUES IN ROUTER 24-5

Computer Networks
 Network Performance-1
• Packet delay versus network load
• Delay is composed of propagation delay and processing delay

Computer Networks 24-6

 Network Performance-2
• Throughput versus network load
• Throughput: the number of packets passing through the network in a unit of time

Computer Networks 24-7

 Congestion Control
• Congestion control refers to techniques and mechanisms that can either
prevent congestion, before it happens, or remove congestion, after it has
happened.

• Two broad categories: open-loop congestion control (prevention) and

closed-loop congestion control (removal).

Computer Networks 24-8

 Open Loop Control: Prevention
• Retransmission policy and timers must to be designed to
optimize efficiency and at the same time prevent congestion

• Window policy: Selective Repeat is better than Go-back-N

• Acknowledgement policy: does not ACK every packet

• Discard policy: prevent congestion and at the same time may

not harm the integrity of the transmission

• Admission policy: Switch first check the resource requirement

of a flow before admitting it to the network

Computer Networks 24-9

Closed-Loop Congestion Control: Removal
• Back pressure: inform the previous upstream router to reduce
the rate of outgoing packets if congested

• Choke point: a packet sent by a router to the source to inform

it of congestion, similar to ICMP’s source quench packet

Computer Networks 24-10

Closed-Loop Congestion Control: Removal
• Implicit signaling: slow down its sending rate by
detecting an implicit signal concerning congestion

• Explicit signaling: Backward signaling / Forward

signaling

Computer Networks 24-11

 Congestion Control in TCP
• TCP assumes that the cause of a lost segment is due to congestion in the
network.
• If the cause of the lost segment is congestion, retransmission of the
segment does not remove the cause—it aggravates it.
• The sender has two pieces of information: the receiver-advertised window
size and the congestion window size
• TCP Congestion window
– Actual window size = minimum (rwnd, cwnd)
(where rwnd = receiver window size, cwnd = congestion window
size)

Computer Networks 24-12

 TCP Congestion Policy
• Based on three phases: slow start, congestion avoidance, and
congestion detection
• Slow Start: Exponential Increase
– In the slow-start algorithm, the size of the congestion window increases
exponentially until it reaches a threshold

Computer Networks 24-13

 TCP Congestion Policy
• Congestion Avoidance: Additive Increase
– The size of the congestion window increases additively until
congestion is detected

Computer Networks 24-14

 TCP Congestion Policy

• Congestion Detection: Multiplicative Decrease

• An implementation reacts to congestion detection in
one of two ways:
– If detection is by time-out, a new slow start phase starts
– If detection is by three ACKs, a new congestion avoidance
phase starts

Computer Networks 24-15

 TCP Congestion Policy
Summary

Computer Networks 24-16

Congestion Example

Computer Networks 24-17

 Congestion Control: Frame Relay
• Congestion avoidance: BECN and FECN
• Backward explicit congestion notification (BECN)
• Forward explicit congestion notification (FECN)

Computer Networks 24-18

Four Cases of Congestion

Computer Networks 24-19

 Quality of Service (QoS)
We can define QoS as something a flow seeks to attain
• Flow Characteristics:
– Reliability
– Delay
– Jitter: the variation in delay for packets belonging to the same flow
– Bandwidth

• Flow Classes:
– Based on the characteristics, we can classify flows into groups,
with each group having similar levels of characteristics
Computer Networks 24-20
 Techniques to improve QoS
1. Scheduling:
i. FIFO queuing
ii. Priority queuing
iii. Weighted fair queuing
2. Traffic shaping:
i. Leaky bucket
ii. Token bucket
3. Resource reservation: buffer, bandwidth, CPU, etc..
4. Admission control: accept or reject a flow based on
predefined parameters called flow specification.
Computer Networks 24-21
 Techniques to improve QoS
• FIFO queuing

Computer Networks 24-22

 Priority Queuing
• Packets are first assigned to priority class. Each priority class
has its own queue
• The packets in the highest-priority queue are processed first

Computer Networks 24-23

 Weighted Fair Queuing
• The queues are weighted based on the priority of the queues
• The system processes packets in each queue in a round-robin fashion
with the number of packets selected from each queue based on the
weight.

Computer Networks 24-24

 Traffic Shaping: Leaky Bucket
• Traffic shaping: to control the amount and the rate of the traffic sent to
network
• A leaky bucket algorithm shapes bursty traffic into fixed-rate traffic
by averaging the data rate. It may drop the packets if the bucket is full.

Computer Networks 24-25

 Leaky Bucket Implementation

• Algorithm for variable-length packets:

1) Initialize a counter to n at the tick of the clock
2) If n is greater than the size of the packet, send packet and
decrement the counter by the packet size. Repeat this step
until n is smaller than the packet size
3) Reset the counter and go to step 1
Computer Networks 24-26
 Traffic Shaping: Token Bucket
• The token bucket allows bursty traffic at a regulated maximum rate.

• Token bucket + leaky bucket: leaky bucket after token bucket

Computer Networks 24-27

Resource Reservation

A flow of data needs resources such as a buffer, bandwidth, CPU

time, and so on. The quality of service is improved if these resour
ces are reserved beforehand.

Admission Control

Admission control refers to the mechanism used by a router, or a

switch, to accept or reject a flow based on predefined parameters
called flow specifications. Before a router accepts a flow for proc
essing, it checks the flow specifications to see if its capacity (in
terms of bandwidth, buffer size, CPU speed, etc.) and its previous
commitments to other flows can handle the new flow.
Computer Networks 24-28
Two models have been designed to provide quality of service in
the Internet:

1. Integrated Services
2. Differentiated Services

Both models emphasize the use of quality of service at the network

layer (IP).

Computer Networks 24-29

 Integrated Services (IntServ)
• Integrated Services is a flow-based QoS model designed for IP, which means the
user needs to create a flow, a kind of virtual circuit from the source to the
destination and inform all routers of the resource requirements.
• Signaling: Resource Reservation Protocol (RSVP)
• Flow specification: for reservation
– Rspec (resource specification) defines the resource that the flow needs to
reserve
– Tspec (traffic specification) defines the traffic characterization of the flow

• Admission: a router decides to admit or deny the flow specification

• Service classes: guaranteed service and controlled-load service
– Guaranteed service class: guaranteed minimum end-to-end delay
– Controlled-load service class: accept some delays, but is sensitive to an
overloaded network and to the danger of losing packets
Computer Networks 24-30
 RSVP: Resource reSerVation Protocol

• In IntServ, the resource reservation is for a flow, a kind of virtual

circuit network out of the IP
• RSVP is a signaling protocol to help IP create a flow and
consequently make a resource reservation
• RSVP is a signaling system designed for multicasting (multicat
trees)
• Receiver-based reservation
• RSVP message: Path and Resv
• Path message: from sender to all receivers
• Resv message: from receivers to sender

Computer Networks 24-31

RSVP: Resource reSerVation Protocol

Computer Networks 24-32

 Resv Messages
• Make a resource reservation from each receiver to sender

Reservation Merging

Computer Networks 24-33

 Reservation Styles
• When there is more than one flow, the router needs to make a
reservation to accommodate all of them.
• Wild card filter style: a single reservation for all senders
• Fixed filter style: a distinct reservation for each flow
• Shared explicit style: a single reservation which can be shared by a
set of flow

Computer Networks 24-34

Soft State

• The reservation information (state) stored in every node for a

flow needs to be refreshed periodically. This is referred to as a
soft state.
• The default interval for refreshing is currently 30 s.

Problems with Integrated Services

1. Scalability
2. Service-type limitation

Computer Networks 24-35

 Differentiated Service (Diffserv)
• Differentiated Services is a class-based QoS model designed for IP.
• Diffserv handles the shortcomings of IntServ
• Main differences between Diffserv and Intserv
– Main processing is moved from the core to the edge (scalability)
– The per-flow is changed to per-class flow service (service-type
limitation)
• DS field
– DSCP (DS Code Point) is a 6-bit field that define per-hop behavior
(PHB)
– CU (currently unused) is 2-bit

Computer Networks 24-36

 Per-hop Behavior (PHB)

• Diffserv defines three PHBs

• DE PHB (default PHB) is the same as best-effort delivery
• EF PHB (expedited forwarding PHB) provides the following
services:
– Low loss, low latency, ensured bandwidth
• AF PHB (assured forwarding PHB) delivers the packet with a
high assurance as long as the class traffic does not exceed the
traffic profile of the node

Computer Networks 24-37

 Traffic Conditioner
• Meter checks to see if the incoming flow matches the negotiated traffic
profile
• Marker can re-mark a packet with best-effort delivery or down-mark a
packet based on the meter information; no up-mark
• Shaper use the meter information to reshape the traffic if not compliant
with the negotiated profile.
• Dropper, like a shaper with no buffer, discard packets if the flow severely
violates the profile

Computer Networks 24-38

 QoS in Switched Network
• QoS in Frame Relay
– Four different attributes are used to control traffic
– Access rate,
– committed burst size (Bc),
– committed information rate (CIR) = Bc/T bps
– excess burst size (Be)

Computer Networks 24-39

 User Rate in Relation to Bc and Bc + Be
• How can a user send bursty data ?

Computer Networks 24-40

 QoS in ATM
• QoS in ATM is based on the class, user related attributes, and
network-related attributes
• Classes: CBR, VBR, ABR, and UBR
– CBR (constant): real-time audio or video over dedicated T-line
– VBR (variable): compressed audio or video, VBR-RT, VBR-NRT
– ABR (available): bursty application
– UBR (unspecified): best-effort delivery

Computer Networks 24-41

 QoS in ATM

• User-related attributes: define how fast the user wants to send data
– SCR (sustained cell rate): average cell rate over a long time
interval
– PCR (peak cell rate)
– MCR (minimum cell rate)
– CVDT (cell variation delay tolerance)

• Network-related attributes: define characteristics of the network

– CLR (cell loss ratio) = lost / sent
– CTD (cell transfer delay)
– CDV (cell delay variation)
– CER (cell error ratio)
Computer Networks 24-42