delays in packet networks
DESCRIPTION
Delays in Packet Networks. Getting a Grip on. Jorg Liebeherr Almut Burchard Dept. of ECE Dept. of Mathematics University of Toronto. Packet Switch. Fixed-capacity links Variable delay due to waiting time in buffers Delay depends on Traffic Scheduling. Traffic Arrivals. - PowerPoint PPT PresentationTRANSCRIPT
![Page 1: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/1.jpg)
Delays in Packet NetworksDelays in Packet Networks
Jorg Liebeherr Almut Burchard Dept. of ECE Dept. of Mathematics
University of Toronto
Getting a Grip on Getting a Grip on
![Page 2: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/2.jpg)
Packet SwitchPacket Switch
• Fixed-capacity links
• Variable delay due to waiting time in buffers
• Delay depends on
1. Traffic
2. Scheduling
![Page 3: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/3.jpg)
Traffic ArrivalsTraffic Arrivals
MPEG-Compressed Video Trace
0
50
100
150
200
250
300
350
0 200 400 600 800 1000
Frame number
Tra
ffic
(in
50
byt
e ce
lls)
Peak rate
Mean rate
Fram
e s
ize
Frame number
![Page 4: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/4.jpg)
First-In-First-OutFirst-In-First-Out
![Page 5: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/5.jpg)
Static Priority (SP)Static Priority (SP)
• Blind Multiplexing (BMux):
All “other traffic” has higher priority
![Page 6: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/6.jpg)
Earliest Deadline First (EDF)Earliest Deadline First (EDF)
Benchmark scheduling algorithm for meeting delay requirements
![Page 7: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/7.jpg)
NetworkNetwork
![Page 8: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/8.jpg)
Simplified NetworkSimplified Network
...
Through flow
Crosstraffic
Crosstraffic
Crosstraffic
Computing delays in such networks is notoriously hard …
![Page 9: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/9.jpg)
Over the last (almost) 20/15 years, we have worked on problems relating to network delays:
• Worst-case delays
• Scheduling vs. statistical multiplexing
• Statistical bounds on end-to-end delays
• Difficult traffic types
• Scaling laws
… … but temptingbut tempting
![Page 10: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/10.jpg)
CollaboratorsCollaborators
• Domenico Ferrari
• Dallas Wrege
• Hui Zhang
• Ed Knightly
• Robert Boorstyn
• Chaiwat Oottamakorn
• Stephen Patek
• Chengzhi Li
• Florin Ciucu
• Yashar Ghiassi-Farrokhfal
![Page 11: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/11.jpg)
• J.Liebeherr, D. E. Wrege, D. Ferrari, “Exact admission control for networks with a bounded delay service,” ACM/IEEE Trans. Netw. 4(6), 1996.
• E. W. Knightly, D. E. Wrege, H. Zhang, J. Liebeherr, “Fundamental Limits and Tradeoffs of Providing Deterministic Guarantees to VBR Video Traffic,” ACM Sigmetrics, 1995.
• R. Boorstyn, A. Burchard, J. Liebeherr, C. Oottamakorn. “Statistical Service Assurances for Packet Scheduling Algorithms”, IEEE JSAC, Dec. 2000.
• A. Burchard, J. Liebeherr, S. D. Patek, “A Min-Plus Calculus for End-to-end Statistical Service Guarantees,” IEEE Trans. on IT, 52(9), Sep. 2006.
• F. Ciucu, A. Burchard, J. Liebeherr, “A Network Service Curve Approach for the Stochastic Analysis of Networks”, ACM Sigmetrics 2005.
• C. Li, A. Burchard, J. Liebeherr, “A Network Calculus with Effective Bandwidth,” ACM/IEEE Trans. on Networking, Dec. 2007.
• J. Liebeherr, A. Burchard, F. Ciucu, “Non-asymptotic Delay Bounds for Networks with Heavy-Tailed Traffic,” INFOCOM 2010.
• J. Liebeherr, Y. Ghiassi-Farrokhfal, A. Burchard, “On the Impact of Link Scheduling on End-to-End Delays in Large Networks,” IEEE JSAC, May 2011.
• J. Liebeherr, Y. Ghiassi-Farrokhfal, A. Burchard, “The Impact of Link Scheduling on Long Paths: Statistical Analysis and Optimal Bounds”, INFOCOM ’2011.,
• A. Burchard, J. Liebeherr, F. Ciucu, “On Superlinear Scaling of Network Delays, ACM/IEEE Trans. Netw., to appear.
Papers Papers (relevant to this talk)(relevant to this talk)
![Page 12: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/12.jpg)
DisclaimerDisclaimer
• This talk makes a few simplifications
• Please see papers for complete details
![Page 13: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/13.jpg)
Traffic DescriptionTraffic Description
• Traffic arrivals in time interval [s,t) is
• Burstiness can be reduced by “shaping” traffic
Arrival function A
![Page 14: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/14.jpg)
Regulatedarrivals
Flows areshaped
Buffered Link
Flow 1
Flow N )(NE
C
€
A1
.
.
.
€
AN
Traffic is shaped by an envelope such that:
Shaped ArrivalsShaped Arrivals
)(1 E
s
P
Popular envelope: “token bucket”
![Page 15: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/15.jpg)
• Link capacity C
• Each flows j has
• arrival function Aj
• envelope Ej
• delay requirement dj
What is the maximum number of shaped flows with delay requirements that can be put on a single buffered link?
![Page 16: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/16.jpg)
Delay Analysis of SchedulersDelay Analysis of Schedulers
•Consider arrival from flow i at t with t+di:
•Tagged arrival departs by if
€
supy
Aj(t − y, t + Δ ij ) −C(y + di)j
∑ ⎧ ⎨ ⎪
⎩ ⎪
⎫ ⎬ ⎪
⎭ ⎪≤ 0
Arrivals from flow j
Taggedarrival
t
•Consider a link scheduler with rate C
Deadline ofTagged arrival
Limit(Scheduler Dependent)
ijt yt
![Page 17: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/17.jpg)
yt t
Arrivals from flow j
with
FIFO:
Static Priority:
EDF:
.0 ij
€
ij = −∞ (lower) , 0 (same) , di (higher).
jiij dd
Delay Analysis of SchedulersDelay Analysis of Schedulers
€
supy
Aj(t − y, t + Δ ij ) −Cyj
∑ ⎧ ⎨ ⎪
⎩ ⎪
⎫ ⎬ ⎪
⎭ ⎪≤ Cdi
ijt
![Page 18: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/18.jpg)
Schedulability ConditionSchedulability Condition
We have:
., )(),( tEttA jj
An arrival from class i never has a delay bound violation if
ij
ijjy
CdCyyE
)(sup
Therefore:
Condition is tight, when Ej is concave
![Page 19: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/19.jpg)
C = 45 Mbps
MPEG 1 traces:
Lecture:d = 30 msec
Movie (Jurassic Park):d = 50 msec
Type 1 flows
strong effectiveenvelopes
Numerical Result Numerical Result (Sigmetrics 1995)(Sigmetrics 1995)
EDF
Static Priority (SP)Peak Rate
![Page 20: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/20.jpg)
Deterministicworst-case
Expected case
Probable worst-case
![Page 21: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/21.jpg)
Statistical Multiplexing GainStatistical Multiplexing Gain
Flow 1Arrivals Flow 2
Flow 3
Time
Worst-case arrivalsB
ack
log
Worst-casebacklog
Flow 1Flow 2Flow 3
Time
Bac
klo
g
Arrivals
With statistical multiplexing
Backlog
![Page 22: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/22.jpg)
Statistical Multiplexing GainStatistical Multiplexing Gain
flow 1 for
guarantees
support to
needed Resources
N
flows N for
guarantees
support to
needed Resources
Statistical multiplexing gain is the raison d’être for packet networks.
![Page 23: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/23.jpg)
What is the maximum number of flows with delay requirements that can be put on a buffered link and considering statistical multiplexing?
Arrivals are random processes
• Stationarity: The are stationary random processes
• Independence: The and are stochastically independent
![Page 24: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/24.jpg)
Envelopes for random arrivalsEnvelopes for random arrivals
• Statistical envelopeStatistical envelope :
• Statistical sample path envelopeStatistical sample path envelope :
Statistical envelopes are non-random functions
Statistical envelope bounds arrival from flow j with high certainty
![Page 25: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/25.jpg)
Aggregating arrivalsAggregating arrivals
Arrivals from group of flows:
with deterministic envelopes: with deterministic envelopes:
with statistical envelopes:with statistical envelopes:
![Page 26: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/26.jpg)
Statistical envelope for group of Statistical envelope for group of indepenent (shaped) flows indepenent (shaped) flows
• Exploit independence and extract statistical multiplexing gain when calculating
• For example, using the Chernoff Bound, we can obtain
![Page 27: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/27.jpg)
Statistical vs. Statistical vs. Deterministic Deterministic Envelope Envelope EnvelopesEnvelopes
Type 1 flows:P =1.5 Mbps = .15 Mbps =95400 bits
Type 2 flows:P = 6 Mbps = .15 Mbps = 10345 bits
Type 1 flows
statisticalenvelopes
(JSAC 2000)(JSAC 2000)
![Page 28: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/28.jpg)
Traffic rate at t = 50 msType 1 flows
Statistical vs. Statistical vs. Deterministic Deterministic Envelope Envelope Envelopes Envelopes (JSAC 2000)(JSAC 2000)
![Page 29: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/29.jpg)
Deterministic ServiceNever a delay bound violation if:
Scheduling AlgorithmsScheduling Algorithms
• Work-conserving scheduler that serves Q classes
• Class-q has delay bound dq
-scheduling algorithm
Scheduler
)(1 sE
€
EQ (s)
.
.
.
Statistical ServiceDelay bound violation with if:
C
![Page 30: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/30.jpg)
Statistical Multiplexing vs. Scheduling Statistical Multiplexing vs. Scheduling (JSAC 2000)(JSAC 2000)
Statistical multiplexing makes a big difference
Scheduling has small impact
Example: MPEG videos with delay constraints at C= 622 Mbps Deterministic service vs. statistical service ( = 10-6)
Thick lines: EDF SchedulingDashed lines: SP scheduling
dterminator=100 ms dlamb=10 ms
![Page 31: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/31.jpg)
More interesting traffic typesMore interesting traffic types
• So far: Traffic of each flow was shaped
• Next:
• On-Off traffic
• Fraction Brownian Motion (FBM) traffic
Approach: • Exploit literature on
Effective Bandwidth • Derived for many traffic
types
MPEG-Compressed Video Trace
0
50
100
150
200
250
300
350
0 200 400 600 800 1000
Frame number
Tra
ffic
(in
50
byt
e ce
lls)
Peak rate
Mean rate
effectivebandwidth
![Page 32: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/32.jpg)
Statistical Envelopes and Effective Statistical Envelopes and Effective Bandwidth Bandwidth
Effective Bandwidth (Kelly 1996)
Given , an effective envelope is given by
![Page 33: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/33.jpg)
Comparisons of statistical service guarantees for different schedulers and traffic types
Schedulers:
SP- Static PriorityEDF – Earliest Deadline FirstGPS – Generalized Processor Sharing
Traffic:
Regulated – leaky bucketOn-Off – On-off sourceFBM – Fractional Brownian Motion
C= 100 Mbps, = 10-6
Different Traffic Types Different Traffic Types (ToN 2007)(ToN 2007)
![Page 34: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/34.jpg)
Delays on a path with multiple nodes:
• Impact of Statistical Multiplexing
• Role of Scheduling • How do delays scale with path length?
• Does scheduling still matter in a large network?
![Page 35: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/35.jpg)
• Systems theory for networksin (min,+) algebra
developed by Rene Cruz, C. S. Chang, JY LeBoudec (1990’s)
• Service curve S characterizes node
• Used to obtain worst-case bounds on delay and backlog
.....
........
............
.
D(t)
A(t)
s
..........
..
backlog=B(s)
delay=W(s)
Deterministic Network Calculus Deterministic Network Calculus (1/3)(1/3)
A Darrivals departures
Node
S
![Page 36: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/36.jpg)
Deterministic Network Calculus Deterministic Network Calculus (2/3)(2/3)
• Worst-case view of
• arrivals:
• service :
• Implies worst-case bounds
• backlog:
• delay :
• (min,+) algebra operators
• Convolution:
• Deconvolution:
A Darrivals departures
Node
S
![Page 37: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/37.jpg)
Deterministic Network Calculus Deterministic Network Calculus (3/3)(3/3)
• Main result:
If describes the service at each node, then
describes the service given by the network as a whole.
Sender ReceiverSender Receiver
![Page 38: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/38.jpg)
Stochastic Network CalculusStochastic Network Calculus
• Probabilistic view on arrivals and service
• Statistical Sample Path Envelope
• Statistical Service Curve
• Results on performance bounds carry over, e.g.:
• Backlog Bound
![Page 39: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/39.jpg)
Stochastic Network CalculusStochastic Network Calculus
• Hard problem: Find so that
• Technical difficulty:
is a random variable!
![Page 40: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/40.jpg)
Statistical Network Service Curve Statistical Network Service Curve ((Sigmetrics Sigmetrics 2005)2005)
• Notation:
• Theorem: If are statistical service curves, then for any :
is a statistical network service curve with some finite violation probability.
![Page 41: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/41.jpg)
EBB modelEBB model
• Traffic with Exponentially Bounded Burstiness (EBB)
• Sample path statistical envelope obtained via union bound
![Page 42: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/42.jpg)
Example: Scaling of Delay BoundsExample: Scaling of Delay Bounds
• Traffic is Markov Modulated On-Off Traffic (EBB model)
• All links have capacity C
• Same cross-traffic (not independent!) at each node
• Through flow has lower priority:
...
CrossFlows
CrossFlows
CrossFlows
CrossFlows
CrossFlows
CrossFlows
ThroughFlows
ThroughFlows
Node HNode 2Node 1
![Page 43: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/43.jpg)
Example: Scaling of Delay BoundsExample: Scaling of Delay Bounds
• Two methods to compute delay bounds:
1. Add per-node bounds: Compute delay bounds at each node and sum up
2. Network service curve: Compute single-node delay bound with statistical network service curve
...
CrossFlows
CrossFlows
CrossFlows
CrossFlows
CrossFlows
CrossFlows
ThroughFlows
ThroughFlows
Node HNode 2Node 1
![Page 44: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/44.jpg)
Example: Scaling of Delay Bounds Example: Scaling of Delay Bounds (Sigmetrics (Sigmetrics 2005)2005)
• Peak rate: P = 1.5 MbpsAverage rate: = 0.15 Mbps
• T= 1/ + 1/ = 10 msec
• C = 100 Mbps• Cross traffic = through traffic • = 10-9
• Addition of per-node bounds grows O(H3)
• Network service curve bounds grow O(H log H)
![Page 45: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/45.jpg)
• M/M/1 queues with identical exponential service at each node
Result: Lower Bound on E2E Delay Result: Lower Bound on E2E Delay (Infocom (Infocom 2007)2007)
Theorem: E2E delay satisfies for all
Corollary: -quantile of satisfies
![Page 46: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/46.jpg)
Numerical examplesNumerical examples
• Tandem network without cross traffic
• Node capacity:
• Arrivals are compound Poisson process
• Packets arrival rate:
• Packet size:
• Utilization:
...
Through flow
Node HNode 2Node 1
![Page 47: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/47.jpg)
Upper and Lower Bounds on E2E Delays Upper and Lower Bounds on E2E Delays (Infocom (Infocom 2007)2007)
![Page 48: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/48.jpg)
Superlinear Scaling of Network Superlinear Scaling of Network Delays Delays
• For traffic satisfying “Exponential Bounded Burstiness”, E2E delays follow a scaling law of
• This is different than predicted by
… worst-case analysis
… networks satisfying “Kleinrock’s independence assumption”
![Page 49: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/49.jpg)
Back to scheduling … Back to scheduling …
So far: Through traffic has lowest priority and gets leftover capacity
Leftover Service
or Blind Multiplexing
BMux C
How do end-to-end delay bounds look like for different schedulers?
Does link scheduling matter on long paths?
![Page 50: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/50.jpg)
Service curves vs. schedulers Service curves vs. schedulers (JSAC 2011)(JSAC 2011)
• How well can a service curve describe a scheduler?
• For schedulers considered earlier, the following is ideal:
with indicator function and parameter
![Page 51: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/51.jpg)
Example: End-to-End BoundsExample: End-to-End Bounds
• Traffic is Markov Modulated On-Off Traffic (EBB model)
• Fixed capacity link
...
CrossFlows
CrossFlows
CrossFlows
CrossFlows
CrossFlows
CrossFlows
ThroughFlows
ThroughFlows
Node HNode 2Node 1
![Page 52: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/52.jpg)
Example: Deterministic E2E Delays Example: Deterministic E2E Delays (Infocom ‘11)(Infocom ‘11)
• Peak rate: P = 1.5 MbpsAverage rate: = 0.15 Mbps
• C = 100 Mbps
BMUX
EDF(delay-tolerant)
FIFO
EDF(delay intolerant
![Page 53: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/53.jpg)
Example: Statistical E2E Delays Example: Statistical E2E Delays (Infocom`11)(Infocom`11)
• Peak rate: P = 1.5 MbpsAverage rate: = 0.15 Mbps
• C = 100 Mbps• = 10-9
![Page 54: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/54.jpg)
Example: Statistical Output Burstiness Example: Statistical Output Burstiness (Infocom ‘11)(Infocom ‘11)
• Peak rate: P = 1.5 MbpsAverage rate: = 0.15 Mbps
• C = 100 Mbps• = 10-9
![Page 55: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/55.jpg)
Can we compute scaling of delays for nasty traffic ?
![Page 56: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/56.jpg)
• A heavy-tailed process satisfies
with
• A self-similar process satisfies
Hurst Parameter
Heavy-Tailed Self-Similar TrafficHeavy-Tailed Self-Similar Traffic
![Page 57: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/57.jpg)
• Heavy-tailed self-similar (htss) envelope:
• Main difficulty: Backlog and delay bounds require sample path envelopes of the form
• Key contribution (not shown): Derive sample path bound for htss traffic
htts Traffic Envelopehtts Traffic Envelope
![Page 58: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/58.jpg)
Example: Node with Pareto Traffic Example: Node with Pareto Traffic (Infocom (Infocom
2010)2010)
Traffic parameters:
Node:• Capacity C=100 Mbps
with packetizer• No cross traffic
Compared with:• Lower bound from
Infocom 2007• Simulations
![Page 59: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/59.jpg)
Example: Nodes with Pareto Traffic (End-to-Example: Nodes with Pareto Traffic (End-to-end)end)
Parameters:
Compared with:• Lower bound from
Infocom 2007
• Simulation traces of 108 packets
![Page 60: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/60.jpg)
Upper Bound:
Lower Bound:
Bounds:
Illustration of scaling boundsIllustration of scaling bounds (Infocom 2010)(Infocom 2010)
(N)
Lower Bound
(N log N)
Upper Bound
![Page 61: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/61.jpg)
Summary of insights Summary of insights
① Satisfying delay bounds does not require peak rate allocation for complex traffic
② Statistical multiplexing gain dominates gain due to link scheduling
③ scaling law of end-to-end delays
④ New laws for heavy-tailed traffic
⑤ Link scheduling plays a role on long path
1995 2000 2005 2010
(1) (2) (3) (4), (5)
![Page 62: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/62.jpg)
Example: Pareto TrafficExample: Pareto Traffic
• Size of i–th arrival:
• Arrivals are evenly spaced with gap :
• With Generalized Central Limit Theorem …
… and tail bound
• ... we get htss envelope
-stable distribution
![Page 63: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/63.jpg)
Example: Envelopes for Pareto Traffic Example: Envelopes for Pareto Traffic (Infocom (Infocom 2010)2010)
Parameters:
Comparison of envelopes:
• htss GCLT envelope
• Average rate
• Trace-based• deterministic
envelope
• htts trace envelope
![Page 64: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/64.jpg)
Single Node Delay BoundSingle Node Delay Bound
• htss envelope:
• ht service curve:
• Delay bound:
![Page 65: Delays in Packet Networks](https://reader036.vdocument.in/reader036/viewer/2022062322/568148f5550346895db61598/html5/thumbnails/65.jpg)
ConclusionsConclusions
Requirements
Queueing networks
Effective bandwidth
Network calculus
Traffic classes (incl. self-similar, heavy-tailed)
Limited Broad Broad(but loose)
Scheduling Limited No Yes
QoS (bounds on loss, throughput delay)
Very limited
Loss, throughput
Deterministic
Statistical Multiplexing
Some Yes No
Stochastic network calculus
Broad
Yes
Yes
Yes