Điện tử viễn thông lecture8 congestion control khotailieu

ATM Traffic Control Schemes Reactive controls: Adaptive admission control Call routing and load balancing Adaptive usage parameter control Explicit congestion notification forward o

Congestion Control and Traffic Management

in High Speed Networks

Carey Williamson

University of Calgary

Introduction

 Mục đích của điều khiển tắc nghẽn la

 Điều chỉnh luồng lưu lượng trong mạng

 Để tránh lam bão hòa hoặc quá tải các node trung

gian

Congestion: Effects

 Tắc nghẽn có thể gây ra:

Gia tăng trê, do xếp hang trong mạng Packet loss, do tran buffer

Reduced throughput, do mất packet va

truyền lại

 Tương tự: giao thông “rush hour”

 Tuyến đầu ra chậm hơn đầu vao

 Các nguồn lưu lượng cạnh tranh cho cùng đầu ra tại cùng một thời điểm

Buffering: A Solution?

 Ghi đệm tại các switches có thể giúp giảm

bớt trình trạng tắc nghẽn ngắn hạn hoặc nhất

thời

 Tuy nhiên, nếu quả tải liên tục, buffer sẽ bị

đầy nhanh chóng, va packets sẽ bị lost

 Chỉ trì hoãn vấn đề tắc nghẽn

 Tăng buffer đồng nghĩa với trễ xếp hang lớn hơn

 Vượt quá một điểm giới hạn, tăng buffer lam cho

Động cơ (tiếp)

 Yêu cầu buffer tăng theo tốc độ tuyến

VD, to store 1 second worth of traffic @ 64 kbps: 8 kilobytes

@ 10 Mbps: 1.25 Mbytes

@ 1 Gbps: 125 Mbytes

Động cơ (tiếp)

 Tính không đồng nhất của tốc độ các tuyến

- Bổ sung các tuyến tốc độ cao vao mạng

không có nghĩa la các tuyến tốc độ thấp cũ sẽ mất đi

- Kết nối các mạng tốc độ cao va thấp tạo nên tắc nghẽn tại các điểm kết nối

Động cơ (tiếp)

 Lưu lượng rất bursty

- high peak-to-mean ratio, peak rates

- e.g., data traffic: 10-to-1, 1-10 Mbps

- e.g., video traffic: 20-to-1, 5-100 Mbps

- có thể statistically multiplex một số kênh nhưng nếu nhiều kênh cùng hoạt động thì không thể tránh khỏi tắc nghẽn

Motivation (Cont’d)

 Thời gian phản ứng được giới hạn bởi trễ

lan truyền

- Trong mạng WAN tốc độ cao tích

Trễ-Băng thông lớn!!!

- Tích Trễ-Băng thông cho biết có bao

nhiêu bít trong đường truyền giữa hai đầu

- Vao thời điểm nhận biết được mạng bị tắc nghẽn thì đã có hang Mbit hay nhiều hơn dữ liệu được gửi vao mạng!!

Giải pháp phòng chống tắc

 Phát hiện tắc nghẽn, hoặc tấn công vao tắc

nghẽn, hanh động để giải quyết vấn đề trước khi tình trạng tắc nghẽn trở nên xấu hơn

 Preventive: reservation-based

 Ngăn tắc nghẽn khỏi xảy ra bằng cách đặt chỗ

Reactive vs Preventive

 Phần lớn các giải pháp của Internet là các

cơ cấu reactive

 TCP Slow Start

 Random-Early-Detection (RED) Gateways

 Tích trễ-băng thông lớn lam cho nhiều giải pháp không thể áp dụng được cho các

mạng tốc độ cao

 Phần lớn các giải pháp điều khiển tắc

nghẽn của ATM là preventive,

reservation-based

Congestion Control in ATM

 Congestion control in high speed ATM networks

 Call-level controls

 Cell-level controls

Call-Level Control (Cont’d)

 Yêu cầu tai nguyên tại thời điểm thiết lập cuộc gọi (connection establishment) cho thời gian của cuộc gọi (vd, bandwidth,

buffers)

 Nếu tai nguyên cho phép, cuộc gọi được thực hiện

 Nếu không, cuộc gọi bị chặn lại

 Ví dụ, tín hiệu bận trong telephone network

Call-Level Control (Cont’d)

 Tradeoff: QoS va Utilization

 Utilization: nhiều cuộc gọi, hiệu suât cao

 QoS: nhiều cuộc gọi, xác suất tắc nghẽn cao

Call-Level Control (Cont’d)

 Nhược điểm:

Có thể không công bằng

- denial of service, long access delay

Khó xác định chính xác yêu cầu tai nguyên

va các tham số QoS

- có thể không biết hoặc không trung thực

- tắc nghẽn vẫn có thể xảy ra

Cell-Level Control

 Còn gọi la input rate control

 Điều khiển tốc độ đầu vao của nguồn lưu

lượng để ngăn ngừa, giảm, hoặc điều khiển mức độ tắc nghẽn

 Có thể có nhiều cơ cấu:

Traffic shaping, traffic policing, UPC

Leaky bucket (token bucket)

Cell tagging (colouring), cell discarding

Cell scheduling disciplines

VD: Congestion Control in ATM

 Có đầy đủ các giải pháp điều khiển lưu lượng từ ngắn hạn (e.g., traffic shaping, cell

discarding) đến dai hạn (e.g., network

provisioning)

ATM Traffic Control Schemes

Time Scale

ATM Traffic Control Schemes

Short

usec

ATM Traffic Control Schemes

ATM Traffic Control Schemes

Long

ATM Traffic Control Schemes

ATM Traffic Control Schemes

ATM Traffic Control Schemes

Time Scale

Call

Duration

Admission Control Routing, Load Balancing

ATM Traffic Control Schemes

Long

ATM Traffic Control Schemes

Long

ATM Traffic Control Schemes

 Preventive controls:

Resource provisioning

Connection admission control

Call routing and load balancing

Usage parameter control

Priority control

Traffic shaping

Fast reservation protocol

ATM Traffic Control Schemes

 Reactive controls:

Adaptive admission control

Call routing and load balancing

Adaptive usage parameter control Explicit congestion notification

(forward or backward)

Node to node flow control

Selective cell discarding

Leaky Bucket

 One of the cell-level control mechanisms that

has been proposed is the leaky bucket (a.k.a token bucket)

 Has been proposed as a traffic policing

mechanism for Usage Parameter Control

(UPC), to check conformance of a source to its traffic descriptor

Can also be used as a traffic shaper

Leaky Bucket (Cont’d)

 Think of a bucket (pail) with a small hole in the bottom

 You fill the bucket with water

 Water drips out the bottom at a nice constant rate: drip, drip, drip

Leaky Bucket (Cont’d)

Leaky Bucket (Cont’d)

Bucket

Leaky Bucket (Cont’d)

Bucket

Empty

Leaky Bucket (Cont’d)

Bucket

Hole

Leaky Bucket (Cont’d)

Bucket

Water

Hole

Leaky Bucket (Cont’d)

Leaky Bucket (Cont’d)

Drip

Leaky Bucket (Cont’d)

Leaky Bucket (Cont’d)

Leaky Bucket (Cont’d)

Leaky Bucket (Cont’d)

Constant rate stream of drips, all nicely spaced, periodic

Leaky Bucket (Cont’d)

Constant rate stream of drips, all nicely spaced, periodic

Storage area

for drips

waiting to go

Leaky Bucket (Cont’d)

 A leaky bucket flow control mechanism is then a software realization of this very

simple idea

 Packets (cells) waiting for transmission

arrive according to some (perhaps

unknown) arrival distribution

 Tokens arrive periodically (deterministically)

 Cell must have a token to enter network

Leaky Bucket (Cont’d)

Leaky Bucket (Cont’d)

3 4

5

Leaky Bucket (Cont’d)

Incoming Cells

Incoming Tokens

1 2

3 4

5

Leaky Bucket (Cont’d)

Incoming Cells

Incoming Tokens

+ 1 ToNetwork

2 3 4

5

Leaky Bucket (Cont’d)

Incoming Cells

Incoming Tokens

+ 2 1 ToNetwork

3 4

5

Leaky Bucket (Cont’d)

Leaky Bucket (Cont’d)

Leaky Bucket (Cont’d)

Leaky Bucket (Cont’d)

Leaky Bucket (Cont’d)

 Cells must obtain tokens in order to proceed into the network

 If no token available, then cell is discarded

 Constrains the rate at which cells enter the network to be the rate negotiated at the time

of call setup

 Shapes traffic, reduces burstiness

Buffered Leaky Bucket

 Arriving cells that find a token waiting can proceed directly into the network

 Arriving cells that find no token ready must wait in queue for a token

 Cells that arrive to a full queue are lost

 Tokens that arrive to a full token pool are simply discarded

Buffered Leaky Bucket

Buffered Leaky Bucket

Output traffic is r cells/sec, nicely paced

 If X < r, then tokens always ready

Output traffic is X (< r)

 Use for traffic shaping or UPC

Buffered Leaky Bucket

(Cont’d)

 A station can “save up” at most M tokens

 Limits the maximum burst size in the network

 Can send at most M cells back to back

 B can be set to balance the tradeoff between cell loss and cell delay

Leaky Bucket UPC

 The token rate r is set based on the rate

declared at the time of call setup

 Makes sure that each source obeys rate

that was used when the call admission

decision was made (i.e., descriptor)

 Can use “single leaky bucket” to police just the peak cell rate (PCR)

 Can use “dual leaky bucket” to police both PCR and SCR

 There are several different variations of the basic leaky bucket concept described in the literature, such as the virtual leaky bucket,

spacer, others

 Basic idea: rather than strictly enforcing

rates, allow senders to occasionally exceed their prescribed rate, as long as they mark or tag their extra cells

Cell Marking Scheme

 Uses leaky bucket to regulate cell

transmissions as before, but rather than

having cells wait for tokens when there are

no tokens ready, the station can transmit the cell and mark it as a violation cell (i.e., cell colouring)

 Green (CLP = 0) for cells that obey rate

 Red (CLP = 1) for cells that don’t

Cell Colouring (Cont’d)

 If the network detects congestion at any

point, then it does not hesitate to throw away

red cells (CLP = 1), but always gives

preference to green cells

 Green cells must get through

 Red cells get through only if there is spare capacity in the network

Selective Cell Discard (SCD)

 A cell-level control mechanism in ATM

switches called selective cell discard can be implemented quite easily using a CLP

threshold on each queue/buffer

 Below the threshold, can accept both green

and red cells

 Beyond the threshold, can only accept green

cells

Selective Cell Discard (Cont’d)

Buffer in an ATM switch

Selective Cell Discard (Cont’d)

Buffer in an ATM switch

Selective Cell Discard (Cont’d)

Buffer in an ATM switch

Some cells waiting to go

Selective Cell Discard (Cont’d)

CLP Threshold

Buffer in an ATM switch

Selective Cell Discard (Cont’d)

CLP Threshold

Selective Cell Discard (Cont’d)

CLP Threshold

Selective Cell Discard (Cont’d)

CLP Threshold

Selective Cell Discard (Cont’d)

CLP Threshold

Selective Cell Discard (Cont’d)

CLP Threshold

Selective Cell Discard (Cont’d)

CLP Threshold

Selective Cell Discard (Cont’d)

CLP Threshold

Selective Cell Discard (Cont’d)

CLP Threshold

Selective Cell Discard (Cont’d)

CLP Threshold

Selective Cell Discard (Cont’d)

CLP Threshold

Selective Cell Discard (Cont’d)

CLP Threshold

Selective Cell Discard (Cont’d)

CLP Threshold

Selective Cell Discard (Cont’d)

CLP Threshold

Selective Cell Discard (Cont’d)

CLP Threshold

Selective Cell Discard (Cont’d)

CLP Threshold

Selective Cell Discard (Cont’d)

CLP Threshold

Selective Cell Discard (Cont’d)

CLP Threshold

Selective Cell Discard (Cont’d)

CLP Threshold

Selective Cell Discard (Cont’d)

CLP Threshold

Selective Cell Discard (Cont’d)

CLP Threshold

Explicit Congestion Notification

 There are some proposals to use reactive congestion control approaches for end-to-end flow control in ATM

 One of the mechanisms proposed is called Explicit Forward Congestion Notification

(EFCN) (or EFCI, for Explicit Forward Congestion Indication)

EFCI: Basic Operation

 Switches can detect the onset of congestion (e.g., buffers filling up)

 Switches set a control bit in cell headers to indicate this congestion condition

 Sources react by reducing the volume of

traffic that they are sending through that

switch

 Suitable for VBR or ABR traffic

EFCI: Basic Operation (Cont’d)

Traffic

Source

Traffic Sink

EFCI: Basic Operation (Cont’d)

Traffic

Source

Traffic Sink

Buffer

EFCI: Basic Operation (Cont’d)

Traffic

Source

Traffic Sink

EFCI: Basic Operation (Cont’d)

Traffic

Source

Traffic Sink

EFCI Threshold

EFCI: Basic Operation (Cont’d)

Traffic

Source

Traffic Sink

Data Cell

EFCI: Basic Operation (Cont’d)

Traffic

Source

Traffic Sink

EFCI Threshold

EFCI: Basic Operation (Cont’d)

Traffic

Source

Traffic Sink

EFCI: Basic Operation (Cont’d)

Traffic

Source

Traffic Sink

EFCI Threshold

EFCI: Basic Operation (Cont’d)

Traffic

Source

Traffic Sink

EFCI: Basic Operation (Cont’d)

Traffic

Source

Traffic Sink

EFCI Threshold

EFCI: Basic Operation (Cont’d)

Traffic

Source

Traffic Sink

EFCI: Basic Operation (Cont’d)

Traffic

Source

Traffic Sink

EFCI Threshold

EFCI: Basic Operation (Cont’d)

Traffic

Source

Traffic Sink

!!!

EFCI: Basic Operation (Cont’d)

Traffic

Source

Traffic Sink

EFCI Threshold

Ack Cell

EFCI: Basic Operation (Cont’d)

Traffic

Source

Traffic Sink

EFCI: Basic Operation (Cont’d)

Traffic

Source

Traffic Sink

EFCI Threshold

EFCI: Basic Operation (Cont’d)

Traffic

Source

Traffic Sink

EFCI: Basic Operation (Cont’d)

Traffic

Source

Traffic Sink

EFCI Threshold

!!!

EFCI: Basic Operation (Cont’d)

Traffic

Source

Traffic Sink

EFCI: Basic Operation (Cont’d)

Traffic

Source

Traffic Sink

EFCI Threshold

EFCI: Basic Operation (Cont’d)

Traffic

Source

Traffic Sink

EFCI: Basic Operation (Cont’d)

Traffic

Source

Traffic Sink

EFCI Threshold

EFCI: Basic Operation (Cont’d)

Traffic

Source

Traffic Sink

EFCI Issues

 How to set EFCI threshold

 What should sources do when EFCI signal

is seen

 What should sources do when no EFCI

signal is seen

 Forward versus backward notification

 Effect of feedback delay

 Delay x bandwidth product

 Congestion control in high speed ATM

networks is a difficult problem

 Lots of good ideas of how to do it, but no real standard (yet?)

 Will likely require a combination of schemes

at different time scales and for different

classes of traffic