on internet traffic dynamics and internet topology ii ... · pdf fileon internet traffic...
TRANSCRIPT
![Page 1: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/1.jpg)
On Internet Traffic Dynamics and Internet Topology II
Internet Model Validation
Walter WillingerAT&T Labs-Research
![Page 2: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/2.jpg)
Topics Covered
§ Self-similar network traffic§ Mandelbrot’s legacy§ Self-similarity through high variability
§ A new model validation framework§ Illustrations of validation framework§ Self-similar traffic vs. Poisson§ Chaos and self-similarity§ SOC and the Internet§ The “New Science of Networks” and the Internet
§ Lessons learned
![Page 3: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/3.jpg)
![Page 4: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/4.jpg)
![Page 5: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/5.jpg)
![Page 6: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/6.jpg)
Time scale: 100 sec
10 msec
Time scale factor: 10 at each step
Time scale: 100 sec
10 msec
Time s
cale f
actor:
10 at
each s
tep
![Page 7: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/7.jpg)
Time scale: 100 sec
10 msec
Varibility scales like 10 (̂-0.8) at each step
Time scale: 100 sec
10 msec
Variabi
lity sc
ales li
ke 10
(̂-0.
5) at e
ach ste
p
![Page 8: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/8.jpg)
Mandelbrot’s Construction
§ Renewal reward processes and their aggregates§ Aggregate is made up of many constituents§ Each constituent is of the on/off type§ On/off periods have a “duration” § Constituents make contributions (“rewards”) when “on”§ Constituents make no contributions when “off”
§ What can be said about the aggregate?§ In terms of assumed type of “randomness” for durations
and rewards§ In terms of implied type of “burstiness”
![Page 9: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/9.jpg)
A Closely Related Construction
§ Cox’s construction§ Aggregate traffic is made up of many connections§ Connections arrive at random§ Each connection has a “size” (number of packets)§ Each connection transmits packets at some “rate”
§ The limiting regimes for the aggregate are essentially the same as those for Mandelbrot’s construction § Also called immigration-death or M/G/∞ process
![Page 10: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/10.jpg)
Mandelbrot’s Types of “Randomness”
§ Distribution functions/random variables§ “Mild” ? finite variance (Gaussian)§ “Wild” ? infinite variance (heavy tails)
§ Correlation function of stochastic process§ “Mild” ? short-range dependence (SRD, Markovian)§ “Wild” ? long-range dependence (LRD)
![Page 11: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/11.jpg)
Mandelbrot’s Types of “Burstiness”
burstysmooth
BURSTYBursty
Dis
trib
utio
n fu
nctio
n
Mild
Wild
SRD LRD
Correlation structure
• Tail-driven burstiness (“Noah effect”)• Dependence-driven burstiness (“Joseph effect”)
![Page 12: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/12.jpg)
Type of Burstiness: “Smooth”W h i t e N o i s e
Correlation Function r(n)
lag n on linear scale
r(n)
on
log
scal
e
CCDF Function 1-F(x)
x on linear scale
1-F(
x) o
n lo
g sc
ale
Log-linear scales
![Page 13: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/13.jpg)
Type of Burstiness: “bursty”C o l o r e d N o i s e
Correlation Function r(n)
lag n on log scale
r(n)
on
log
scal
e
CCDF Function 1-F(x)
x on linear scale
1-F(
x) o
n lo
g sc
ale
Log-linear scale
Log-log scale
![Page 14: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/14.jpg)
Type of Burstiness: “Bursty”S t a b l e N o i s e
Correlation Function r(n)
lag n on linear scale
r(n)
on
log
scal
e
CCDF Function 1-F(x)
x on log scale
1-F(
x) o
n lo
g sc
ale
Log-log scale
Log-linear scale
?
![Page 15: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/15.jpg)
C o l o r e d S t a b l e N o i s e
Type of Burstiness: “BURSTY”CCDF Function 1-F(x)
x on log scale
1-F(
x) o
n lo
g sc
ale
Correlation Function r(n)
lag n on log scale
r(n)
on
log
scal
e
?
?
Log-log scales
![Page 16: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/16.jpg)
Mathematical Framework
§ Mathematical abstraction§ Ignores most transport-specific features§ Assumes unlimited link bandwidth
§ Consider (normalized) double-limit process§ Large M (many users), large T (coarse time scale)§ Large M limit, then large T limit: “TCP-type” behavior§ Large T limit, then large M limit: “UDP-type” behavior§
§ with the aggregated total workload from 0 to u
[ ])()(
)()(lim , MgTf
TtAETtA MM
TM
−∞→
)(uA M
![Page 17: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/17.jpg)
Mathematical Result (1)
§ Assumed “randomness”§ Durations are heavy-tailed with infinite variance§ Rewards are 0/1
§ Large M-limit, then large T-limit (TCP-like transport)§ “Mild” marginals (Gaussian)§ “Wild” correlations (LRD)
§ Limit process§ Fractional Brownian motion§ Aggregate of many long, thin flows
![Page 18: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/18.jpg)
Web servers
application traffic
is streamed out on the net
creating Internettraffic
Webclient
Let’s look at Internet traffic over some link
![Page 19: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/19.jpg)
Link
Internet traffic
![Page 20: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/20.jpg)
Web traffic …
… is streamed onto the Internet …
… creating “bursty-looking” link traffic
time
X = “wild” HTTP Requests/responses
Y = “colored” noise(TCP-type transport)
![Page 21: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/21.jpg)
Mathematical Result (2)
§ Assumed “randomness”§ Durations are heavy-tailed with infinite variance§ Rewards are 0/1
§ Large T-limit, then large M-limit (UDP-like transport)§ “Wild” marginals (stable)§ “Mild” correlations (independence)
§ Limit process§ Levy stable motion§ Aggregate of many short, fat flows
![Page 22: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/22.jpg)
Web traffic …
… is streamed onto the Internet …
… creating “bursty-looking” link traffic
time
X = “wild” HTTP requests/responses
Y = “stable” noise(UDP-type transport)
![Page 23: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/23.jpg)
A New Model Validation Approach
§ Traditional framework for model validation§ Model “fits” or “describes” the data “well”§ “Well” according to a goodness-of-fit criterion§ Model results in comparable performance
§ Need for a paradigm shift§ New mathematical constructions ensure good “fit”§ Standard goodness-of-fit criteria perform poorly with
“voluminous” or “unusual” data (LRD, heavy tails)§ Performance may be comparable, but wrong (e.g., due
to neglecting crucial feedback regulation)
![Page 24: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/24.jpg)
The “Closing the Loop” Approach
1. Discovery (data-driven)2. Proposed explanation in terms of elementary
concepts or mechanisms (mathematics)3. Step 2 suggests first-of-its-kind measurements or
revisiting existing measurements related to checking the elementary concepts or mechanisms
4. Empirical validation of elementary concepts or mechanisms using the data collected in Step 3
![Page 25: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/25.jpg)
Why “Closing the Loop” is Progress
§ Departure from classical “data-fitting”§ Validation is moved to a more elementary or
fundamental level§ Fully exploits the context in which measurements
are made (“start with data, end with data”)§ If successful, provides actual explanation of
“emergent” phenomena (new insight)§ Shows inherent limitations and weaknesses of
proposed model, suggests further improvements
![Page 26: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/26.jpg)
Illustration of the “Closing the Loop” Approach§ Self-similarity through high variability§ Self-similar traffic vs. Poisson§ Chaos and self-similarity§ Self-organized criticality (SOC) and the
Internet§ The “New Science of Networks” and the
Internet
![Page 27: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/27.jpg)
Example 0: Self-Similarity – High Variability§ Discovery: Self-similar nature of aggregate
traffic on a link§ Mathematics: Mandelbrot/Cox construction§ Suggested new measurements§ Session/flow/connection sizes
§ Analysis of new measurements§ Check session/flow/connection size distributions
![Page 28: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/28.jpg)
Theory-based IP Flow Property
1 0-1
1 00
1 01
1 02
1 03
1 04
1 0-6
1 0-5
1 0-4
1 0-3
1 0-2
1 0-1
1 00
x
1-F
(x)
C o m p l e m e n t a r y C u m u l a t i v e D i s t r i b u t i o n P l o t
P a r e t o ( i n f i n i t e v a r i a n c e )E x p o n e n t i a l ( f i n i t e v a r i a n c e )
![Page 29: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/29.jpg)
Measured IP Flow Sizes
![Page 30: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/30.jpg)
Example 1:Poisson vs. Self-Similar
§ Lucent Press Release, June 6, 2001http://www.lucent.com/press/0601/010606.bla.htlm
§ Cao, Cleveland, Lin, and Sun, ”Internet traffic tends toward Poisson and Independent as the Load Increases” (2002)
![Page 31: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/31.jpg)
Claims and Supporting Evidence
§ Claims§ Internet traffic tends to Poisson and
independent as the load increases§ Packet interarrival times (of aggregate traffic)
become exponentially distributed and independent as the link load increases
§ Supporting evidence§ 2526 packet traces, 5 min or 90 sec in duration§ Impressive-looking statistical analysis of traces
![Page 32: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/32.jpg)
Bottom-line
§ Specious claims based on impressive-looking, but specious analysis of the data§ Specious nature of the claims are revealed by a
straightforward application of the “closing the loop” concept§ Reminder (J.W. Tukey, 1975)
“Often the purpose of good analysis is not so much to do well in catching what you want but rather to well … in rejecting what you don’t want.”
![Page 33: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/33.jpg)
Example 2:Chaos and Self-Similarity
§ Infocom 2000, Best Paper Award§ http://www.ieeeinfocom.org/2000/papers/74.pdf
§ Veres and Boda,”The chaotic nature of TCP congestion control” (2000)
![Page 34: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/34.jpg)
Claims and Supporting Evidence
§ Claims§ TCP itself, as a deterministic process, creates
chaos, which generates self-similarity§ Approach taken§ Simulation study based on ns§ Chaos theory-inspired analysis of simulated
trace data (phase plots, attractors, Lyapunov exponent)§ Ad-hoc testing of self-similarity
![Page 35: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/35.jpg)
Bottom-line
§ Specious claim as a result of a specious experiment/simulation/analysis§ Considerable, not-so-solid follow-up work§ ns can generate pseudo-chaotic observations§ Reminder (Floyd and Paxson, 2001)
“Network simulators make it easier for researchers to create simulations, but the researchers themselves remain responsible for making their use of simulation relevant and insightful, rather than irrelevant or misleading.”
![Page 36: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/36.jpg)
Example 3: Self-Organized Criticality (SOC) and the Internet§ Csabi, “1/f noise in computer network traffic”
(1994)§ Ohira and Sawatari, “Phase transition in computer
traffic models” (1998)§ Yuan, Ren, and Shan, “Self-organized criticality
in a computer network model” (2000)§ Sole and Valverde, “Information transfer and
phase transitions in a model of Internet traffic” (2001)
![Page 37: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/37.jpg)
Claims and Supporting Evidence
§ (Selected) Claims§ Self-similar traffic traffic is the result of a
phase transition§ There exist deep links between Internet traffic
and highway traffic§ Approach§ Simulation-based studies (cellular automata)§ Ignores Internet-specific “details” for the sake
of addressing “fundamental” issues
![Page 38: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/38.jpg)
Bottom-line
§ Specious claims as a result of a specious model § Specious nature of the claims is revealed by a
trivial “closing the loop” argument§ Building a network with the claimed properties
seems feasible but is not recommended!§ Reminder (Floyd and Paxson, 2001)
“Simulation carries with it the risk of using a model simplified to the point where key facets of Internet behavior have been lost, in which case any ensuing results could be useless (though they may not appear to be so.”
![Page 39: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/39.jpg)
Example 4: The “New Science of Networks” and the Internet
§ Nature, July 2000, “Achilles’ heel of the Internet’’§ Albert, Jeong, and Barabasi, “Error and attack
tolerance of complex networks” (2000)§ A.-L. Barabsi, “Linked: The New Science of
Networks” (2002)
![Page 40: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/40.jpg)
Claims and Supporting Evidence
§ (Selected) Claims§ The Internet topology is “scale-free”§ The Internet topology is robust to random attack, but
highly vulnerable to targeted attacks (“Achilles’ heel”)
§ Approach taken§ Develop new class of “scale-free” graphs§ Model Internet topology as “scale-free” graph§ Simulation studies with “scale-free” graphs
![Page 41: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/41.jpg)
Bottom-line
§ Specious claims as a result of specious models for Internet connectivity§ Shows the power of the “closing the loop” concept§ Building an Internet with “scale-free” properties
may be possible, but is not recommended!§ The “New Science of Networks” has little
networking relevance (so far!)
![Page 42: On Internet Traffic Dynamics and Internet Topology II ... · PDF fileOn Internet Traffic Dynamics and Internet Topology II ... §Sole and Valverde, “Information transfer and](https://reader031.vdocuments.site/reader031/viewer/2022021818/5ab7eeee7f8b9ad5338c2d95/html5/thumbnails/42.jpg)
Lesson Learned
§ Expect specious claims and theories about the Internet [“specious” – having deceptive attraction or allure, or having a false look of truth orgenuiness]§ Press releases (URL’s), accepted INFOCOM papers
(even if they get the Best Paper Award), published articles in Nature or Science (especially on Internet-related topics), etc. can be wrong!