Differentiated Services

2007

Two Approaches for ProvidingQoS on the Internet

“Freeway model” -- integrated services Internet (intserv)– Build a dedicated highway or “circuit” between

communicating points (VIP) “Doctor’s model” -- differentiated services (diffser

v)– Mark a doctor’s vehicle (e. g.,ambulance) or “p

acket” to get priority the road and limit the percentage of such high- priority vehicles in the total traffic mix (fire-engine, policeman)

Difficulties with Intserv and RSVP• Intsev

– A connection-oriented solution– QoS on a per-flow basis– Depends on resource reservation (RSVP) - signaling

• Scalability– Keeping a state (and using it!) for each flow ov

erloads the routers– Periodic messages to refresh the states create

more traffic• Router complexity increases • How to satisfy heterogeneous QoS requirements f

or different receivers• Ambitious signaling is not practical

Differentiated Services (Diff-Serv)• Motivations

– Don’t want end-to-end signaling and per-flow state, for scalability, complexity and quick-to-deploy reasons

• Connectionless mode• QoS for aggregates of traffic, such as premium, assured a

nd best- effort – Avoid strong assumptions about traffic types

• Keep the forwarding path simple– Push complexity to network edge -> Differentiated

Services Domain (DS Domain)• Per-Hop-Behavior (PHB)

– Define behavior of individual routers rather than end-to-end services

– forwarding behavior

Differentiated Service (DS) Field• Diff-Serv Approach: Use the TOS field to sort packets into cl

asses and routers treat them differently– e.g., delay requirement, drop precedence etc.

• DS filed reuse the first 6 bits from the TOS byte– Diff- Service Code Point (DSCP)

• The other two bits are proposed to be used by ECN

Version HLen TOS Length

Identification Fragment offsetFlags

Source address

Destination address

TTL Protocol Header checksum

0 4 8 16 19 31

Data

IPheader

DS Filed0 5 6 7

Differentiated Services (Diffserv)• Build around the concept of domain• Domain – a contiguous region of network under th

e same administrative ownership• Differentiate between edge and core routers• Edge routers

– Perform per aggregate shaping or policing– Mark packets with a small number of bits; each

bit encoding represents a class (subclass)• Core routers

– Process packets based on packet marking• Far more scalable than Intserv, but provides weak

er services

Diffserv Architecture• Ingress routers

– Police/shape traffic

– Set Diff- Service Code Point (DSCP) in Diffserv (DS) field

• Core routers

– Implement Per Hop Behavior (PHB) for each DSCP

– Process packets based on DSCP

IngressEgressEgress

IngressEgressEgress

DS-1 DS-2

Edge router Core router

Differentiated Services• Two types of QoS service

– Two PHBs defined so farPremium service

• EF - Expedited forwarding (type P)Assured service

• AF - Assured forwarding (type A)– 4 classes, each guaranteed a minimum

amount of bandwidth and buffering– each with three drop preference partit

ions• Plus, best-effort service

Premium Service• EF - Expedited Forwarding• Provides guaranteed peak bandwidth service with

negligible delay / jitter– Provides the abstraction of a virtual pipe

between an ingress and an egress router• Network: guarantees that premium packets are

not dropped and they experience low delay with requested profile

• User: sends within profile. User does not send more than the size of the pipe– If it sends more, excess traffic is delayed, and

dropped when buffer overflows• Rate limiting of EF packets at edges only, using

token bucket to shape transmission

Assured Service (1)• AF -- Assured Forwarding• Provides an expected level of bandwidth with delay• Permits flows to use any additional available

bandwidth

• Network: provides lower loss rate than best-effort– In case of congestion best-effort packets are

dropped first• User: sends no more assured traffic than its profile

– If it sends more, the excess traffic is converted to best-effort

Assured Service (2)• User and network agree to some traffic profile

(how much traffic is a user allowed to send)– Edges mark packets up to allowed rate as “in-

profile” or low drop precedence – Other packets are marked with higher drop

precedence • A congested DS node tries to protect packets with a

lower drop precedence value from being lost by preferably discarding packets with a higher drop precedence value– Implemented using buffer management, e. g., RIO --

RED (random early detection) with In and Out

Assured Service (3)• Large spatial granularity service• Theoretically, user profile is defined irrespective of

destination– All other services we learnt are end-to-end, i.e.,

we know destination(s) apriority• This makes service very useful, but hard to

provision (why ?)

Ingress

Traffic profile

Components of Diff-Serv(1) Edge node algorithms

– Classification, policing, shaping, metering, marking, etc.

(2) Router algorithms

– Packet discard algorithms: RED-like methods for in-profile and out-of-profile traffic

– Priority, low-latency queueing

Ingress Edge Router

Classifier

Traffic conditioner

Traffic conditioner

Scheduler

Class 1

Class 2

Best-effort

Marked traffic

Ingress

Per aggregateClassification (e.g., user)

Data traffic

Traffic Conditioning

Wait fortoken

Set EF bitPacketinput

Packetoutput

Test iftoken

Set AF “in” bit

token

No token

Packetinput

Packetoutput

Drop on overflow

• Traffic conditioner (TC) implement– Metering– Marking– Shaping

Scheduler

• Employed by both edge and core routers

• For premium service – use strict priority, or weighted fair queuing (WFQ)

• For assured service – use RIO (RED with In and Out)– Has two classes, “In” and “Out” (of profile)– Always drop OUT packets first– For OUT measure entire queue– For IN measure only in-profile queue

RIO (RED with In and Out) Drop Probability

P(drop)

1.0

MaxP

Minin MaxinMaxoutMinout

AvgLen

Scheduler Example

• Premium traffic sent at high priority

• Assured and best-effort traffic pass through RIO and then sent at low priority

P-bit set?

A-bit set? RIO

yes

no

yes

no

high priority

low priority

Core Router Output Processing

What DSCP?

If “in” setincr in_cnt

High-priority Q

Low-priority Q

If “in” setdecr in_cnt

RIO queuemanagement

Packets out

EF

AF

Control Path• Each domain is assigned a Bandwidth Broker

(BB)– Usually, used to perform ingress-egress

bandwidth allocation • BB is responsible to perform admission control in

the entire domain• BB not easy to implement

– Require complete knowledge about domain– Single point of failure, may be performance

bottleneck– Designing BB still a research problem

Example

• Achieve end-to-end bandwidth guarantee

BBBB BBBB BBBB

1

2 3

579

sender

receiver

8 profile 6profile

4 profile

Comparison

Service

Service Scope

Complexity

Scalability

• Connectivity• No isolation• No guarantees

• End-to-end

• No set-up

• Highly scalable• (nodes maintain

only routing state)

Best-Efforts

• Per aggregation isolation

• Per aggregation guarantee

• Domain

• Long term setup

• Scalable (edge routers maintains per aggregate state; core routers per class state)

Diffserv

• Per flow isolation• Per flow guarantee

• End-to-end

• Per flow setup

• Not scalable (each router maintains per flow state)

Intserv

Summary• Diffserv more scalable than Intserv

– Edge routers maintain per aggregate state– Core routers maintain state only for a few traffic

classes

• But, provides weaker services than Intserv– Per aggregate bandwidth guarantee (premium se

rvice) vs. per flow bandwidth and delay guarantee

• BB is not an entirely solved problem– Single point of failure– Handle only long term reservations (hours, days)

Intserv over Diffserv-- Integrated Services Operation over Diffserv Networks

• Intserv enables per-flow end-to-end QoS– enables hosts to request per-flow, quantifi

able resources, along end-to-end data paths and to obtain feedback regarding admissibility of these requests

• Diffserv enables scalability across large networks

• => Intserv over Diffserv – Integrated Services Operation over Diffse

rv Networks

Network Configuration _______ ___________ ______

/ \ / \ / \

/ \ / \ / \

|----| | |-----| |-----| |------| |-----| | |-----|

|Tx |-| |ER1|---|BR1| |BR2|---|ER2| |-|Rx |

|----| | |-----| |-----| |------| |-----| | |-----|

\ / \ / \ /

\________/ \____________/ \_______/

Non-Diffserv region Diffserv region Non-Diffserv region

• ER -- edge routers which are adjacent to the Diffserv region

• BR -- border routers within the Diffserv region

QoS over Next Generation Networks

• Concluding Remark by H. T. Kung 1998:– In 1993 ATM was in, but it was out of favor in 1997– In 1995 IP switching was in, but it was out of favor

in 1997– In 1996 RSVP was in, but it is out of favor in 1998– In 1998 Diff-Serv and MPLS are in, but will they

also be out of favor soon?• IWQoS2000 panel discussion:

– ATM is on its way out.– Intserv is dead.– Diffserv is dying.– MPLS is in misguiding.

• Growing use of Multi-Protocol Label Switching now

Topics

• Integrated and Differentiated Services• Active Buffer Management techniques, ex. RED.• Multiprotocol Label Switching• QoS in 3G wireless multimedia networks• QoS based routing• QoS management in multi-bearer IP networks i

ncluding PSTN• Impact of media compression in QoS networks• QoS over Mobile IP• QoS enabled MAC design