Schedule of Class Meeng
Mul$media Networking
#10 QoS Semester Ganjil 2012 PTIIK Universitas Brawijaya
Schedule of Class Mee$ng
1.Introduc$on 2.
14. Summary
13. Discussion
CDN: Case Studies 11.
CDN: Solu$ons 10.
Overlay Mul$cast 9.
IP Mul$cast (cont’d) 8.
RTP 6. IP Mul$cast 7.
5.
Coding and Compression
Requirements of MN 4.
Applica$ons of MN 3.
QoS on the Internet: Constraints 12
QoS on the Internet: Solu5ons
Today’s Outline
- Network support for mul$media
- – General QoS
- – Differen$ated QoS
- – Per-‐Connec$on Flow QoS
Network support for mul$media
Approach Granularity Guarantee Mechanism Complex? Deployed? all traffic applica$on-‐layer Making best-‐ treated support (CDNs & none, or so` minimal everywhere effort service equally overlays) packet marking, Differen$ated per class of none, or so` policing, medium some service traffic scheduling packet marking,Per-‐ per source-‐ so` or hard, policing, connec$on des$na$on once is high licle scheduling; call
QoS flow admiced admission
Dimensioning best effort networks
- approach: deploy enough link capacity so that
conges$on doesn’t occur, mul$media traffic flows without delay or loss
- – low complexity of network mechanisms (use current “best effort” network)
- – high bandwidth costs
- challenges:
how much bandwidth is “enough?”
- – network dimensioning:
needed to determine how much bandwidth is “enough” (for that much traffic)
- – es3ma3ng network traffic demand:
- thus far: making the best of best effort service
- – one-‐size fits all service model
- alterna$ve: mul$ple classes of service
- – par$$on traffic into classes
- – network treats different classes of traffic differently (analogy: VIP service versus regular service) 0111
granularity: differen$al service among mul$ple classes , not among individual connec$ons history: ToS bits
Mul$ple classes of service: scenario
H3 H1 R1 R2 H4 R1 output
1.5 Mbps link H2 interface queue Scenario 1: mixed HTTP and VoIP
• example: 1Mbps VoIP, HTTP share 1.5 Mbps link.
- – HTTP bursts can congest router, cause audio loss
- – want to give priority to audio over HTTP
R1 R2
Principle 1
packet marking needed for router to distinguish between different classes; and new router policy to treat packets accordingly Principles for QOS guarantees (more)
- what if applica$ons misbehave (VoIP sends higher than declared rate)
- – policing: force source adherence to bandwidth alloca$ons
provide protec$on (isola$on) for one class from others
- marking , policing at network edge
Principle 2
R1 R2
1 Mbps phone packet marking and policing
- alloca$ng fixed (non-‐sharable) bandwidth to flow: inefficient use of bandwidth if flows doesn ’t use its alloca$on
while providing isola$on, it is desirable to use
resources as efficiently as possible Principle 3R1 R2
1 Mbps phone
Principles for QOS guarantees (more) Scheduling and policing mechanisms
- scheduling: choose next packet to send on link
- FIFO (first in first out) scheduling: send in order of
arrival to queue
- – real-‐world example?
if packet arrives to full queue: who to discard?
- – discard policy:
drop arriving packet
- tail drop:
drop/remove on priority basis
- priority:
drop/remove randomly queue (waiting area) packet arrivals packet departures link
- random:
(server) Scheduling policies: priority high priority queue (waiting area) priority scheduling: send highest arrivals departures priority queued packet classify link (server) low priority queue
- mul$ple classes,
(waiting area) with different
2 priori$es
5
4
1
3 arrivals
- – class may depend on marking or other
packet in 1 3 2 4
5 header info, e.g. IP service source/dest, port departures numbers, etc.
1
3
4
2
5
- – real world example?
- mul$ple classes
- cyclically scan class queues, sending one complete packet from each class (if available)
- real world example?
2
5
1
4
3 arrivals packet in
1
3
2
4
5 service departures
1 3 3 4
5 Scheduling policies: s$ll more
Weighted Fair Queuing (WFQ):
- generalized Round Robin
- each class gets weighted amount of service in each cycle
- real-‐world example?
Policing mechanisms
goal: limit traffic to not exceed declared parameters
Three common-‐used criteria:- how many pkts can be
(long term) average rate: sent per unit $me (in the long run)
- – crucial ques$on: what is the interval length: 100 packets per sec or 6000 packets per min have same average!
- e.g., 6000 pkts per min (ppm) avg.;
peak rate: 1500 ppm peak rate
- max number of pkts sent
(max.) burst size: consecu$vely (with no intervening idle)
Policing mechanisms: implementa$on
token bucket: limit input to specified burst size and average rate
- bucket can hold b tokens
• tokens generated at rate r token/sec unless bucket full
- over interval of length t: number of packets admiHed
less than or equal to (r t + b)
guarantee! WFQ token rate, r bucket size, b per-flow rate, R
Policing and QoS guarantees
- token
bucket,
WFQ
combine
to
provide
guaranteed upper bound on delay, i.e., QoS
D = b/R max arriving traffic arriving traffic Differen$ated services
- want “qualita$ve” service classes
- – “behaves like a wire”
- – rela$ve service dis$nc$on: Pla$num, Gold, Silver
scalability: simple func$ons in network core,
rela$vely complex func$ons at edge routers (or
hosts)- – signaling, maintaining per-‐flow router state difficult with large number of flows
- don’t
define
define
service
classes,
provide
func$onal components to build service classes
DiffServ architecture edge router: per-flow traffic management
marks packets as in-profile and
out-profile core router: per class traffic management
buffering and scheduling based on
marking at edge preference given to in-profile
packets over out-of-profile packets
DiffServ architecture marking r edge router: b per-flow traffic management
marks packets as in-profile and
out-profile core router: per class traffic management
buffering and scheduling based on
marking at edge preference given to in-profile
packets over out-of-profile packets
DiffServ architecture edge router: scheduling per-flow traffic management
marks packets as in-profile and
out-profile .
. . core router: per class traffic management
buffering and scheduling based on
marking at edge preference given to in-profile
packets over out-of-profile packets
Edge-‐router packet marking profile: pre-‐nego$ated rate r, bucket size b
packet marking at edge based on per-‐flow profile
rate r b user packets
: possible use of marking
class-‐based marking: packets of different classes marked
differently
intra-‐class marking: conforming por$on of flow marked
differently than non-‐conforming one
DSCP unused Classifica$on, condi$oning may be desirable to limit traffic injec$on rate of some class:
Diffserv packet marking: details
- packet is marked in the Type of Service (TOS) in IPv4, and Traffic Class in IPv6
- 6 bits used for Differen$ated Service Code Point (DSCP)
- – determine PHB that the packet will receive
- – 2 bits currently unused
- user declares traffic profile (e.g., rate, burst size)
- traffic metered, shaped if non-‐conforming
Forwarding Per-‐hop Behavior (PHB)
- PHB result in a different observable
(measurable) forwarding performance behavior
- PHB
does
not
specify
what
mechanisms
to
use
to ensure required PHB performance behavior
- examples:
- – class A gets x% of outgoing link bandwidth over $me intervals of a specified length
- – class A packets leave first before packets from class B
Forwarding PHB
PHBs proposed:- expedited forwarding: pkt departure rate of a
class equals or exceeds specified rate
- – logical link with a minimum guaranteed rate
- assured forwarding: 4 classes of traffic
- – each with three drop preference par$$ons
- basic fact of life: can not support traffic
demands beyond link capacity call admission: flow declares its needs, network may
block call (e.g., busy signal) if it cannot meet needs
Principle 4 R1 R2
1 Mbps phone
1 Mbps phone
QoS guarantee scenario resource reserva3on
- – call setup, signaling (RSVP)
- – traffic, QoS declara$on
– per-‐element admission control
- – call setup, signaling (RSVP)
- – traffic, QoS declara$on
- – per-‐element admission control QoS-sensitive scheduling
(e.g., WFQ)
QoS guarantee scenario resource reserva3on
- – call setup, signaling (RSVP)
- – traffic, QoS declara$on
- – per-‐element admission control
request/ reply QoS-sensitive scheduling
(e.g., WFQ)