Leads to an HourglassShaped Architecture Saamer Akhshabi Constantine Dovrolis Georgia Institute of Technology sakhshabiconstantinegatechedu 1 My coauthor Saamer Akhshabi ID: 309193
Download Presentation The PPT/PDF document "The Evolution of Layered Protocol Stacks" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
The Evolution of Layered Protocol Stacks Leads to an Hourglass-Shaped Architecture
Saamer Akhshabi Constantine DovrolisGeorgia Institute of Technologys.akhshabi,constantine@gatech.edu
1Slide2
My co-author, Saamer Akhshabi
(2nd year PhD student, he could not travel to Toronto)
2Slide3
OutlineMotivationModel descriptionResultsConcluding remarks
3Slide4
Where does this work come from?Ethernet
Twisted PairSMTP
HTTP
Thunderbird
Silverlight
Firefox
PPP
Coaxial
Cable
IPv4
UDP
TCP
RTP
Optical
Fiber
MPlayer
Modeling
study of the
Internet protocol stack (architecture) and
its
evolution
…
…
…
…
4Slide5
Why is it an hourglass?
Why
?
-Random?
-Designed?
-
Emergence
?
Ethernet
Twisted
Pair
SMTP
HTTP
Thunderbird
Silverlight
Firefox
PPP
Coaxial
Cable
IPv4
UDP
TCP
RTP
Optical
Fiber
MPlayer
…
…
…
…
5Slide6
What happens at the “waist” compared to other layers?
Frequent
innovations
Frequent
innovations
Conserved (“ossified”)
Ethernet
Twisted
Pair
SMTP
HTTP
Thunderbird
Silverlight
Firefox
PPP
Coaxial
Cable
IPv4
UDP
TCP
RTP
Optical
Fiber
MPlayer
…
…
…
…
6Slide7
How can a new protocol survive at the waist?
X.25SNAEthernet
Twisted
Pair
SMTP
HTTP
Thunderbird
Silverlight
Firefox
PPP
Coaxial
Cable
IPv4
UDP
TCP
RTP
Optical
Fiber
MPlayer
…
…
…
…
ATM
IPv6
7Slide8
What about “Future Internet” those architectures?
Will these architectures also evolve to an hourglass in few years?How to make them more “evolvable”?So that they can better accommodate innovation?So that no single protocol at the waist “kills” all competitors
NDN
XIA
MpbilityFirst
Nebula
?
8Slide9
OutlineMotivationModel: EvoArch
ResultsConclusions9Slide10
Two DisclaimersEvoArch is only an abstraction of protocol stacksEvoArch
does not capture many practical aspects and protocol-specific or layer-specific semanticsEvoArch is certainly not the only model, or “the correct model”, for the emergence of hourglass-shaped network architecturesModels should be judged based on their assumptions, parsimony and predictions10Slide11
Model description
Protocols as nodes
Protocol dependencies as
edges
Products:
P(u)
Substrates:
S(u)
Layer of u: l
(u)
Layered acyclic network
u
Every layer provides a service
L
1
…
4
3
2
11Slide12
The value of a protocolThe value of a protocol depends on the value of its products
Protocols with valuable products are more valuable example TCP, HTTP
1
1
5
2
5
3
1
1
1
1
1
1
12Slide13
The generality of a layer
EthernetTwisted PairSMTP
HTTP
Thunderbird
Silverlight
Firefox
PPP
Coaxial
Cable
IPv4
UDP
TCP
RTP
Optical
Fiber
MPlayer
As we go higher in the stack:
Protocols become
less general – they offer more specialized services
The
probability that a protocol is used by next-layer’s protocols decreases
13Slide14
Generality as a probability
We introduce a parameter called
generality vector s
s(l) : probability that new node at layer l+1 chooses each node at layer l as substrate
s(l) decreases as we go higher in protocol stack
s
(1) = 0.9
s
(3) = 0.5
s
(L-1) = 0.1
14Slide15
Competition between protocolsTwo protocols at the same layer compete if they offer similar servicesi.e., if they have large overlap in their products
HTTP competes with FTP due to several overlapping productsTCP does not compete with UDP because they have minimal service overlap
TCP
UDP
HTTP
FTP
15Slide16
Modeling competition
If c = 3/5u competes with
q and w
q does not
compete with w
q
u
w
Let C(u) be set of competitors of u
Node w competes with u if
c: competition threshold
16Slide17
When does a protocol “die”?Protocols can become extinct due to competition with other protocols
For example, HTTP services cover the set of services provided by FTPCompetition from HTTP has led to FTP’s demise
HTTP
FTP
17Slide18
Modeling protocol deathsA node u dies if its value is significantly less than the value of its strongest (i.e., maximum value) competitor.
z: mortality parameter
18Slide19
Cascade deathsu is w’s
competitorSuppose that w dies due to competition with u (r=3/7)21
4
2
7
2
1
3
1
1
1
1
1
1
1
1
q
u
w
If a node w dies, its products also die if their only substrate is w. This can lead to
cascade deaths.
19Slide20
Protocol birthsBasic birth processNumber of new nodes at given time is a small fraction of total number of nodes in network at that time.
New nodes assigned randomly to layersDeath-regulated birth processThe birth rate at a layer is regulated by the death rate in that layerDiscussed later20Slide21
Summary of EvoArchDiscrete-time modelTime advances in
rounds Each round includesbirth of new nodescompetition among nodes at the same layerpotentially, death of some nodes Key parametersGenerality vector s
Competition threshold cMortality parameter z
21Slide22
OutlineMotivationModel DescriptionResults
Emergence of hourglass structures Controlling the location/width of the waistEvolutionary kernelsProtocol differencesConclusions
22Slide23
Hourglass shapeThe network forms an hourglass structure over time
The waist usually occurs at layer 5 or 6. L = 10
c = 3/5
z
= 1
s(l) = 1-l/L
23Slide24
Hourglass Resemblance Metricw(l) : width of layer l
Minimum occurs at layer bX = {w(l), l = 1, . . . b}
Y = {w(l), l = b, . . .L}
Mann-Kendall
statistic for monotonic trend on the sequences X and
Y: coefficients τ
X and
τ
Y
H = (
τ
Y
–
τ
X
)/
2
H=1 when widths first decrease and then increase (monotonically)
Width
Layer number
w(1)
w(2)
…
w(b)
w(
b+1
)
…
W(L)
24Slide25
RobustnessHigh hourglass scores under a wide range of parameters
25Slide26
Why does EvoArch generatehourglass-shaped networks?
Small generality Low competition (local)Few deathsLarge
generality
High competition
Protocols have similar products
-similar values
Few deaths
Generality close to 50%
Few protocols
with many products
Most other protocols die
26Slide27
How can we get a wider waist?As γ increases
Location of the waist moves to higher layersWidth of waist increases27
γ
is the layer at
which
the generality is 50%
L
ayer
number
γ
0.5
s(l)Slide28
Evolutionary kernels 28Slide29
How can a kernel die?Normalized value of a node: value divided by maximum possible value at that round
If several nodes appear at the next higher layer, andkernel fails to quickly acquire those new possible
products,
s
omeone else may do so..
29Slide30
Death-regulated birth process?What if the birth probability in a layer is regulated by the death probability in that layer
?
30
It
becomes practically
impossible to replace kernels Slide31
What if protocols differ in term of a “quality factor”?The “quality factor” can be interpreted broadly
Performance, Extent of deployment, Reliability or security, Incremental improvements,etc
31Slide32
Effects of quality factorWe still get an hourglassSlower network growthLower part of hourglass is smaller in size
only high quality nodes survive at the lower partKernels are often NOT the highest quality protocols
32Slide33
OutlineMotivationModel DescriptionResults
Concluding remarks33Slide34
What does this mean for the Internet architecture?New way to think about (and teach) Internet’s hourglass architecture
New way to think about “ossification” of protocols at the waistParameterized model for TCP/IP stack: Two protocols compete when their service overlap is more than 70%
A protocol survives only if its value is more than 90% of its strongest competitor’s valueDeath-regulated births
34Slide35
What does this mean for IPv4 vs IPv6?
IPv6 has same products but lower extent of deployment (i.e., lower “quality factor”)IPv6 would find it easier to compete w IPv4 if:It had some distinct products that IPv4 does not haveUnfortunately, it only offers more addressesIPv6 would face easier adoption if it was not presented as “IPv4 replacement” but as “the second network-layer protocol”35Slide36
What does this mean for future Internet architectures?Hourglass structures should be expected if these new architectures evolve/compete
Designers should strive for wider waistMore diverse waist -> more evolvable architectureEvoArch: as the waist moves higher, it also becomes widerHow to push the waist to a higher layer?
See highly relevant paper:L.
Popa, A.
Ghodsi, and I.
Stoica. HTTP as the Narrow Waist of the Future Internet. In
ACM SIGCOMM
HotNets
, 2010
36Slide37
From Networking to Network ScienceHourglass effect in
development of embryosHourglass effect in organization structuresHourglass effect in innate immune system
37