Congestion Control in Data Networks and Internets

1
Chapter 10

• Congestion Control in Data Networks and Internets

2
Introduction

• Congestion occurs when number of packets
  transmitted approaches network capacity
• Objective of congestion control
• keep number of packets below level at which
  performance drops off dramatically

3
Queuing Theory

• Data network is a network of queues
• Packet arrive, they are stored in the input
  buffer of the corresponding port
• The node examines each incoming packet, makes a
  routing decision, moves the packet to the
  appropriate output buffer
• This is statistical time division multiplexing
• If arrival rate gt transmission rate
• then queue size grows without bound and packet
  delay goes to infinity

4
Congestion in Data Networks and Internets

• Congestion the number of packets being
  transmitted through a network begins to approach
  the packet-handling capacity of the network.
• Input and output queues at switch or router

• Any given node has a number of I/O ports attached
  to it.
• There are two buffers at each port, one to accept
  arriving packets, and one to hold packets that
  are waiting to depart.
• Buffer size fixed-size or variable-size
• If packets arrive too fast for the node to
  process them ? input buffer overflow
• If packets arrive too fast than packets can be
  cleared from the outgoing buffers ? output buffer
  overflow

5
At Saturation Point, 2 Strategies

• Discard any incoming packet if no buffer
  available
• Saturated node exercises flow control over
  neighbors
• May cause congestion to propagate throughout
  network

6
Ideal Performance

• I.e., infinite buffers, no overhead for packet
  transmission or congestion control
• Throughput increases with offered load until full
  capacity
• Packet delay increases with offered load
  approaching infinity at full capacity
• Power throughput / delay
• Higher throughput results in higher delay

7
Assumption buffer is infinite size Power
throughput / delay Delay propagation delay (from
SA to DA) processing delay (at each node)
queuing delay (variable delay)
8
Practical Performance

• buffers are finite
• With no congestion control, increased load
  eventually causes moderate congestion throughput
  increases at slower rate than load
• Further increased load causes packet delays to
  increase and eventually throughput to drop to zero

9

• In practice, buffers are finite
• If congestion, the exchange of control signals
  for the congestion consume network capacity
• Point A throughput increases rate is slower than
  the rate of load increase (have control signals)
• Point B buffer overflow, packet discard,
  retransmission packet increase

10
Congestion Control

• Backpressure
• Request from destination to source to reduce rate
• Choke packet ICMP Source Quench
• Implicit congestion signaling
• Source detects congestion from transmission
  delays (ACK packet incurs long delay)
• Packets are discarded (time out occurs)
• ? reduces flow

11
Explicit congestion signaling

• Direction
• Backward
• the congestion avoidance procedures should be
  initiated in the opposite direction of the
  received packet
• Forward
• the congestion avoidance procedures should be
  initiated in the same directions the received
  packet. The information may echo to the source
  station with congestion packet or with window
  size (TCP)
• Categories
• Binary
• a bit is set in a data packet to indicate the
  congestion. The source receives a binary
  congestion indication, it reduces its traffic
  flow
• Credit-based
• like TCP slice window size, the destination
  allows the sender an explicit credit that the
  sender only can send the amount of the credit
  packets
• rate-based
• provide an explicit data-rate limit to the
  source. The source may transmit data at a rate up
  to the set limit. To control congestion, any node
  along the path of the connection can reduce the
  data rate limit in a control message to the
  source.

12
Mechanisms for congestion control
13
Traffic Management

• Design issues related to congestion control
• when a node is saturated and must discard
  packets, the discard policy may consider
  following issues
• Fairness
• Various flows suffer from congestion equally
• Last-in-first-discarded may not be fair
• Quality of Service
• We might wish to treat different traffic flows
  differently
• Voice, video delay sensitive, loss insensitive
• File transfer, mail delay insensitive, loss
  sensitive
• Interactive computing delay and loss sensitive
• Reservations
• One way to avoid congestion and to provide
  assured service to applications is to use a
  reservation scheme
• The network agrees to give a defined QoS so long
  as the traffic flow is within contract parameters
• Policing excess traffic discarded or handled on
  best-effort basis