صفحه 1:
William Stallings
Data and Computer
Communications
Chapter 12
Congestion in
Data Networks
صفحه 2:
What Is Congestion?
I Congestion occurs when the number of
packets being transmitted through the
network approaches the packet handling
capacity of the network
Congestion control aims to keep number of
packets below level at which performance
falls off dramatically
Data network is a network of queues
Generally 80% utilization is critical
Finite queues mean data may be lost
oa
ooo
صفحه 3:
Queues at a Node
صفحه 4:
Effects of Congestion
I Packets arriving are stored at input buffers
0 Routing decision made
0 Packet moves to output buffer
1 Packets queued for output transmitted as fast as
possible
I Statistical time division multiplexing
0 If packets arrive to fast to be routed, or to be
output, buffers will fill
0 Can discard packets
0 Can use flow control
I Can propagate congestion through network
صفحه 5:
Interaction of Queues
صفحه 6:
Ideal 0
Performance ۳
صفحه 7:
Practical Performance
I Ideal assumes infinite buffers and no
overhead
0 Buffers are finite
0 Overheads occur in exchanging
congestion control messages
صفحه 8:
Effects of
Congestion-__
No Control
صفحه 9:
Mechanisms for
Congestion Control
صفحه 10:
Backpressure
I If node becomes congested it can slow down or
halt flow of packets from other nodes
0 May mean that other nodes have to apply
control on incoming packet rates
1 Propagates back to source
0 Can restrict to logical connections generating
most traffic
0 Used in connection oriented that allow hop by
hop congestion control (e.g. X.25)
0 Not used in ATM nor frame relay
0 Only recently developed for IP
صفحه 11:
Choke Packet
0 Control packet
I Generated at congested node
I Sent to source node
I e.g. ICMP source quench
| From router or destination
1 Source cuts back until no more source quench
message
1 Sent for every discarded packet, or anticipated
0 Rather crude mechanism
صفحه 12:
Implicit Congestion Signaling
I Transmission delay may increase with
congestion
0 Packet may be discarded
0 Source can detect these as implicit
indications of congestion
0 Useful on connectionless (datagram)
networks
1١ e.g. IP based
| (TCP includes congestion and flow control - see chapter
17)
0 Used in frame relay LAPF
صفحه 13:
Explicit Congestion Signaling
0 Network alerts end systems of increasing
congestion
0 End systems take steps to reduce offered
load
0 Backwards
I Congestion avoidance in opposite direction to
packet required
0 Forwards
I Congestion avoidance in same direction as
packet required
صفحه 14:
Categories of Explicit Signaling
0 Binary
I A bit set in a packet indicates congestion
0 Credit based
I Indicates how many packets source may send
1 Common for end to end flow control
0 Rate based
I Supply explicit data rate limit
leg. ATM
صفحه 15:
Traffic Management
0 Fairness
0 Quality of service
۱ May want different treatment for different
connections
0 Reservations
I eg. ATM
| Traffic contract between user and network
صفحه 16:
Congestion Control in Packet
Switched Networks
oa
Send control packet to some or all source
nodes
I Requires additional traffic during congestion
Rely on routing information
I May react too quickly
End to end probe packets
I Adds to overhead
Add congestion info to packets as they
cross nodes
I Either backwards or forwards
oa
=
oa
صفحه 17:
ATM Traffic Management
0 High speed, small cell size, limited overhead bits
0 Still evolving
0 Requirements
I Majority of traffic not amenable to flow control
Feedback slow due to reduced transmission time
compared with propagation delay
Wide range of application demands
Different traffic patterns
Different network services
High speed switching and transmission increases
volatility
صفحه 18:
Latency/Speed Effects
0 ATM 150Mbps
0 ~2.8x10- seconds to insert single cell
1 Time to traverse network depends on
propagation delay, switching delay
0 Assume propagation at two-thirds speed of light
1 If source and destination on opposite sides of
USA, propagation time ~ 48x10? seconds
1 Given implicit congestion control, by the time
dropped cell notification has reached source,
7.2x10° bits have been transmitted
1 So, this is not a good strategy for ATM
صفحه 19:
Cell Delay Variation
0 For ATM voice/video, data is a stream of
cells
0 Delay across network must be short
0 Rate of delivery must be constant
0 There will always be some variation in
transit
0 Delay cell delivery to application so that
constant bit rate can be maintained to
application
صفحه 20:
Time Re-assembly of CBR Cells
cells
Successive
صفحه 21:
Network Contribution to
Cell Delay Variation
I Packet switched networks
1 Queuing delays
TI Routing decision time
0 Frame relay
I As above but to lesser extent
0 ATM
Less than frame relay
ATM protocol designed to minimize processing
overheads at switches
ATM switches have very high throughput
Only noticeable delay is from congestion
Must not accept load that causes congestion
صفحه 22:
Cell Delay Variation
At The UNI
0 Application produces data at fixed rate
0 Processing at three layers of ATM causes
delay
I Interleaving cells from different connections
1 Operation and maintenance cell interleaving
I If using synchronous digital hierarchy frames,
these are inserted at physical layer
! Can not predict these delays
صفحه 23:
Origins of Cell Delay Variation
8 octets, X Mbps
(Connection A, X Mbps)
a8 sees ¥ Mbps
AAL (Connection B. Y Mbps)
سس ساسا a= -F fe --- === f+ -F-----F ------F--- ATM layer SAP
زد لا لیس
۳ ۳ ۳ 1
1 1 1
۲ i
ATM Layer 7
۳
oaNE
Physical ۳ ۲ 1 ny a 7 1 11 ۳
Layer
صفحه 24:
Traffic and Congestion
Control Framework
oa
ATM layer traffic and congestion control
should support QoS classes for all
foreseeable network services
Should not rely on AAL protocols that are
network specific, nor higher level
application specific protocols
Should minimize network and end to end
system complexity
=
ص
صفحه 25:
Timings Considered
0 Cell insertion time
0 Round trip propagation time
0 Connection duration
0 Long term
0 Determine whether a given new
connection can be accommodated
0 Agree performance parameters with
subscriber
صفحه 26:
Traffic Management and
Congestion Control Techniques
0 Resource management using virtual paths
0 Connection admission control
0 Usage parameter control
0 Selective cell discard
0 Traffic shaping
صفحه 27:
Resource Management Using
Virtual Paths
I Separate traffic flow according to service
characteristics
User to user application
User to network application
Network to network application
حراج
o
Concern with:
! Cell loss ratio
! Cell transfer delay
! Cell delay variation
صفحه 28:
Configuration of
VCCs and VPCs
صفحه 29:
Allocating VCCs within VPC
0 All VCCs within VPC should experience
similar network performance
0 Options for allocation:
I Aggregate peak demand
I Statistical multiplexing
صفحه 30:
Connection Admission Control
I First line of defence
0 User specifies traffic characteristics for new
connection (VCC or VPC) by selecting a QoS
0 Network accepts connection only if it can
meet the demand
0 Traffic contract
I Peak cell rate
! Cell delay variation
I Sustainable cell rate
I Burst tolerance
صفحه 31:
Usage Parameter Control
I Monitor connection to ensure traffic cinforms
to contract
0 Protection of network resources from
overload by one connection
Done on VCC and VPC
Peak cell rate and cell delay variation
Sustainable cell rate and burst tolerance
Discard cells that do not conform to traffic
contract
0 Called traffic policing
يج نه جر ع
صفحه 32:
Traffic Shaping
0 Smooth out traffic flow and reduce cell
clumping
0 Token bucket
صفحه 33:
Token Bucket
Token Rate p
|
Capacity (3
Departing
Cell
Arriving Capacity K
صفحه 34:
ATM-ABR Traffic Management
1 Some applications (Web, file transfer) do not
have well defined traffic characteristics
0 Best efforts
I Allow these applications to share unused capacity
| If congestion builds, cells are dropped
1 Closed loop control
I ABR connections share available capacity
I Share varies between minimum cell rate (MCR) and
peak cell rate (PCR)
I ARB flow limited to available capacity by feedback
4 Buffers absorb excess traffic during feedback delay
! Low cell loss
صفحه 35:
Feedback Mechanisms
0 Transmission rate characteristics:
I Allowed cell rate
1 Minimum cell rate
I Peak cell rate
I Initial cell rate
0 Start with ACR=ICR
0 Adjust ACR based on feedback from network
I Resource management cells
| Congestion indication bit
4 No increase bit
Explicit cell rate field
صفحه 36:
Variations in Allowed Cell Rate
ime
< ..
pl
13111111۱ 1
1111 ْ
10
MCR
صفحه 37:
Tram
congested,
سس
End Syste
لالم
لانت
Cell Flow
صفحه 38:
Rate Control Feedback
0 EFCI (Explicit forward congestion
indication) marking
0 Relative rate marking
0 Explicit rate marking
صفحه 39:
Frame Relay
Congestion Control
I Minimize discards
0 Miantain agreed QoS
1 Minimize probability of one end user monoply
1 Simple to implement
I Little overhead on network or user
Create minimal additional traffic
Distribute resources fairly
Limit spread of congestion
Operate effectively regardless of traffic flow
Minimum impact on other systems
Minimize variance in QoS
صفحه 40:
Techniques
Discard strategy
Congestion avoidance
Explicit signaling
Congestion recovery
Implicit signaling mechanism
ا
1
1
ا
1
صفحه 41:
Traffic Rate Management
0 Must discard frames to cope with
congestion
I Arbitrarily, no regard for source
| No reward for restraint so end systems transmit
as fast as possible
1 Committed information rate (CIR)
| Data in excess of this liable to discard
0 Not guaranteed
0 Aggregate CIR should not exceed physical data rate
0 Committed burst size
0 Excess burst size
صفحه 42:
Operation of CIR
Commited
Information Current
Rate (CIR) ۵
Maximum
Rate
‘Transmit if
possible
Guaranteed
transmission
Access
rate
صفحه 43:
Relationship
Among ۳
Congestion
Parameters
صفحه 44:
Explicit Signaling
0 Network alerts end systems of growing
congestion
Backward explicit congestion notification
Forward explicit congestion notification
Frame handler monitors its queues
May notify some or all logical connections
User response
I Reduce rate
جم از از —
صفحه 45:
Required Reading
0 Stallings chapter 12