Original Internet Architecture

89 Pages·2016·0.67 MB·English

1
Future Internet Architecture
and challenges of the current Internet
CHIRAG ARORA
S1456717
Master Thesis
Philosophy of Science, Technology and Society
Faculty of Behavioural, Management, and Social Sciences
University of Twente
Supervisors :
Dr. JOHNNYHARTZ SØRAKER &
Dr. MICHAELNAGENBORG
1
2
Acknowledgements
As there are no clear oreven obscure boundaries between my lifeandtimes in writing
thisthesis and my lifeand times in general, thepeople Iam about tothank have had
contributed more tolifethan just helping mein writing thisthesis. Of course, thisputs me
intheprecarious positionofchoosing afew among many.But, perhaps thereader and I
can agree that my failurein acknowledging everyoneis somewhat inevitableand only
trivial,and hopefully Ihave paid my respect to most,ifnot all, whilesharing our
conscious experiences.
With theabovecaveats inmind, Iwould first liketo thank my supervisors Johnny
Soraker andMichael Nagenborg, who have provided mewith good advice but more
importantly,witha lotof fuel for interesting thoughts both beyond and withinthe
classroom.Along with Prof.Ambuj Sagar,who has been afriend, mentorand inspiration
tomefor a numberofyears now,they epitomizewhat Ilikebest aboutAcademiaand
why Isee alife formyself within it.
Being a newcountry,far away from where you were born, can make you think and
re-think themeaning oftheword ‘home’.Thevery fact that I have notformed any clear
opinionsorfound anyreal answers to what that word may mean, is in itselfatestimonyto
allthe friends I havemade in Enschede and Europe in general.Among my ‘international’
friends who really deserve a mention are Jurjen,Thijs, Jannis, Sabrina,Tunmisseand
Jerfy who have lived through my incessant story-telling and my lovefor exploring the
‘other’sidein debates, and who have also played along justas well.Asimilarrolehas
played byVinayak andAteeth, whohavealso madesure that Iam able toget myshare of
exercisein theweek bybikingto a place that isaway from thegrey of thecity.Thetime
spent withthem and thetrees who have witnessed ouraverage poetry recitation skillshas
been as free from thefetters of thepast and future as any timecan be.
Athesis about Internet shouldn't go withoutacknowledging therole ofthistechnology in
2
3
lettingmefeel even among thosewho are physically far away.Most ofthem have been
longenough inmy life toexcludeany possibility ofan acknowledgment that can be
written in afew lines and stillreflect everything Ihave to say.SoI am justgoing to put
outsomenames: Geet,Anuraag, Ritesh,Abhishek, Chetan, Rajat,Ankur,Ashu,Amit,
Parv.Finally,Iwould end thisrather vain exercisebythanking my Parents and my
Brother,who have been great friends and hadan overwhelmingly pleasant role in my
Being.
Let mealso just addthese words byRilke,which reflect what Imay ormay not have
achieved onthispage, orin my thesis, orlifein general, but aspire(d)to:
“...And Iwant mymeaning
trueforyou. Iwantto describe myself
likeapaintingthat Istudied
closelyfor along, longtime,
likeawordIfinallyunderstood,
likethepitcher of water Iuseevery day,
liketheface of mymother,
likeaship
thatcarried me
through thedeadliest stormof all.”
3
4
Abstract
In thelast three decades, theInternet has seen unprecedented growth, scaling from a
few hundredearlyusers to morethan 3billionglobal users,enabling a transformation of
a wide range of economic, social, political and cultural practices. However, along with
this growth, Internet has also witnessed a wide range of growing conflicts among its
stakeholders. The issues raised through these conflicts, which include concerns over
privacy, security, censorship, etc., are at the heart of debates over Internet’s design and
regulation held between stakeholders such as governments, private enterprises, activists
and designers. Given the impact of Internet’s technical architecture over human life, a
number of scholars within the Internet research community, arguing that the current
architecture is too constrained to cater to contemporary concerns, advocate for a
clean-slate architecture design for a future network. Whileaclean-slate architecture
provides us with anopportunity tofind solutions toour current moral dilemmas, italso
leaves us with theresponsibility ofanticipating moral dilemmas that may arise in future
because ofsuch adesign. This thesis aims to take up this responsibility and provide a
step towards a future Internet. Using the conceptual notion of ‘architecture’, which
allows for a decomposition of a complex system, like the Internet, this thesis argues for
four critical meta-requirements for a future Internet architecture. These
meta-requirements, which are properties of the system critical to it beyond its basic
functionalities, cater to four characteristics of a complex system like Internet which
underlie its current as well as potentially future challenges: inherent uncertainty and lack
ofpredictability offuture useof anetwork likeInternet, diverging stakeholder interests
and views, lack offeedback ordata about howthenetwork is being used, and diversity of
societies Internet isembedded in.Iconclude the thesisbyprovidingdesign principles,
which are philosophical guidelines,fora future Internet, along withevaluation tools thata
research team,which I argue shouldbe multi-disciplinary,involved in development ofa
futureInternet can useto evaluate thedifferent specific implementationsof thesedesign
principles and hence, movetowards thedevelopment ofafuture network that can support
thesocio-economic goals ofadiverse world.
4
5
Table of Contents
Chapter1:Introduction....................................................................................................7
Chapter2NetworkArchitecture andSome BasicPrinciples......................................16
2.1System Properties or Meta-Requirements............................................................17
2.3.1Modularity.................................................................................................20
2.3.2Layering....................................................................................................22
2.3.2.1Layering inNetworkArchitectures................................................24
2.3.3End -to-End Principle................................................................................25
2.3.3.2BroadVersion.................................................................................27
Chapter3OriginalInternetArchitecture......................................................................30
3.1Layering...............................................................................................................30
3.1.1Link Layer.................................................................................................30
3.1.2Internet Layer............................................................................................31
3.1.3Transport Layer.........................................................................................31
3.1.4Application Layer......................................................................................32
3.2TheEnd-to-End principle.....................................................................................33
3.3 End-to-End principleand controversies surrounding Internet Policy Guidelines35
Chapter4TheInternet andits Challenges....................................................................38
4.1EconomicChallenges for Internet Service Providers (ISPs)...............................38
4.2Qualityof Service andNetwork Neutrality.........................................................39
4.3Cybersecurity.......................................................................................................41
4.4Emergenceof conflicting viewand goals............................................................43
4.5Extensionsto Internet architecture.......................................................................43
4.6Reflecting ontheproblems ofcurrent Internet forlessons for afuture Network44
Chapter5Reflections forafutureInternet...................................................................46
5.1Uncertainty: Challenges and Opportunities........................................................46
5.2StakeholderConflicts,ValuesandValueSensitiveDesign..................................51
5
6
5.3Network can benefit from added ‘Intelligence’...................................................57
5.4Architecture should beminimally defined...........................................................60
5.5MovingBeyondValueSensitiveDesign(VSD)..................................................60
5.6Meta-requirements for afuture Internet...............................................................63
Chapter6Recommendations forFutureInternet........................................................65
6.1Modularity............................................................................................................65
6.2Design for‘Tussle’...............................................................................................69
6.2.1ModularizealongTussleBoundaries........................................................70
6.2.2Design forChoice (to exertcontrol over tussleoutcomes).......................72
6.3Designing forNetwork Intelligence.....................................................................73
6.3.1FuzzyEnds Principle.................................................................................74
6.3.2Knowledge Plane......................................................................................75
6.4Designing anevolvable, heterogeneous, intelligent Internet fortussle:
Recommendations for Futurework............................................................................77
Chapter7Concluding summary.....................................................................................79
References.........................................................................................................................84
6
7
Chapter 1 : Introduction
In the past three decades, Internet has gone from a small community of users with
mutual trust to one with a wide variety of stakeholders who do not know each other, and
often find themselves having diverging and conflicting interests. For example, many
users want privacy in their communication, while governments want to tap
conversations for a variety of reasons. Certain users want to exchange media, such as
music and movies, with their peers while the media industry, as the legal right holders,
wants to stop them. These conflicts, along with other challenges borne by the increased
size of the Internet, have seen a number of short-term technical fixes that may be
harmful in the long term (Braden et al, 2000). For example, the increased incidences of
security attacks, in the form of hacks or computer viruses, has seen deployment of
Firewalls on computing devices which block potentially (but not always actually)
harmful applications. Similarly, to provide additional services, network providers use
technologies such as Deep Packet Inspection which breach the privacy and transparency
of Internet communication as it looks through the data flowing across the Internet.
These changes violate the Internet’s original design principles and are being widely
debated among scholars, policy makers as well as in forums such as Internet
Engineering Task Force (IETF) (Braden et al, 2000; Van Schewick, 2012:!5). These
debates raise the concerns that such short-term fixes may destroy, or atleast hinder,
Internet’s capability to facilitate innovation, economic growth, free speech, knowledge
exchange, and democratic exchange. Forums such as IETF, and other Standards
Developing Organizations (SDO), have also become important sites for raising and
addressing questions over the moral as well as legal responsibility of technical
designers to protect or enable exercise of ‘human rights-by-design’(Cath and Floridi,
2016) and thus, highlight the growing impact of the technical architecture on human life
(Brown et al. 2010; Clark et al. 2005; Denardis 2013,2014; Lessig,1999).
7
8
As aresponse to thevarious conflicts over Internet’s design, usage andregulation, which
need to be ethically addressed, anumber ofscholars within theInternet research
communityhave suggested a clean-slate architecture design, that is, anew network that is
unconstrained bythecurrent system.Whilea clean-slate architecture provides us with an
opportunityto find solutionsto ourcurrent moral dilemmas, italso leaves us with the
responsibilityofanticipating moral dilemmasthat may arise in future becauseofsuch a
design.Thisresponsibility demands a contribution from ethicists and philosophers, and
thisthesis isdedicated towards this responsibility.
However,before conceptualizing thedesign ofaclean-slate architecture fora complex
network liketheInternet, we require anuanced understanding ofa.) what the term
architecturemeans, as well as howthey are designed inthe contextofnetwork systems,
b.)what is thearchitecture ofthecurrent Internet, c.)howdoes thisarchitecture lead to
theconflicts ofInternet’s design andregulation, d.)what can we learn from these
conflicts and its relation to architecture in orderto design anewarchitecture.The aim of
thisthesis isto address all ofthese issues incorresponding chapter 2-5. In Chapter 6,I
willdraw onthese lessonsand discuss howdesign ofanew architecture can proceed.
Here, Iwill also argue that any research aimed at designing anew network architectureto
match thescale andutility of Internet, should involveamultidisciplinary team which
includes not justtechnical experts but alsonon-technical expertssuch as thosewith
knowledgeofeconomics, social science, philosophy,lawand policy making.
Thecomplexity oflarge systems, such as theInternet, requires thedesigners to
decomposethesystem into smallercomponents.Thearchitecture ofa system describes
howsuch adecomposition is to takeplace as well as howdifferent components interact
witheach otherto produce thedesired functionality.As BarbaraVan Schewick (2012)
defines it, architecture isa “high-level system description that specifies the components
ofthesystem, theexternally visibleproperties of thecomponents, and therelationships
among components”.Different architectures can however,map onto thesame
functionality.Yet, thesearchitectures, with samefunctionality,might differin someother
system properties such as reliability,modifiability,security,and soon.Someofthese
8
9
system properties can beextremely critical to thesocio-economic impact ofthesystem.
Forexample, for asystem present underconditions ofextremeuncertainty,itmight be
advantageous for thesystem to have acapability tobe modified easily.
Designers may often have tomake trade-offs between thedesired system properties as
onesystem property may negatively affect someotherproperty.For example,security as
aproperty may negatively affect asystems capacity to support new applications. Building
security into software systems may require designing specific features to bedesigned into
applications, which may decrease the incentivefor application designers. Similarly,
additionof security features may also block applications originally compatiblewith the
system. One exampleistheuse ofFirewalls on network systems. Firewalls monitorand
filterdata trafficgoing inand out ofacommunication network based onpredefined rules.
Sincedesigning rules to separatebetween malicious and non-malicious datacan an
error-proneexercise, particularly in large networks, firewalls often result in blocking of
non-malicious dataorfunctions (Braden et al, 2000).The decision oftrading offone
system property can therefore becritical in determiningthe behavior ofthesystem, such
as itssocio-economic impacts.In thecontextof thisthesis, which aims to providea step
towards a newnetwork architecture, thisdecision regarding thetrade-offbetween
different critical system properties (also referred to as meta-requirements in this thesis), is
central to thedesign ofanew architecture.
In order tofacilitatethedecision making regarding thetrade-offbetween different
meta-requirements, I discusshowthese meta-requirements were mapped ontothe
architectureof thecurrent Internet, as well as howthat architecture relates to thecurrent
conflicts overInternet’s design and regulation. In the former discussion,I introducethe
concept of‘design principle’,which is used to translate thedesired meta-requirements
ontothesystem architecture. Design principles describe theknown interactions and
outcomes in different architectural choices andguide thedesign of asystem forspecific
meta-requirements. In particular,I discuss threedesign principles, modularity,layering,
and theend-to-end principle and howthey are implemented inthe Internet architecture.
As design principles are conceptual tools, scholars often disagree over theirtechnical
9
10
implementationwithin theInternet architecture.Therefore, in thisthesis, Iwill discuss
and clarify thesedebates.
An understanding ofdesign principles in thecurrent architecture, and thesystem
properties they facilitate, provides afoundation to better understand thedebates
surrounding theconflicts over Internet’s design andregulation, inrelation toits
architecture. In thisthesis, Iaim to unpack theserelations and showhowthecurrent
debates such as thoseover ‘Network Neutrality’or thelack of security measures against
cyber-attacks through hacking and computerviruses can be conceptualizedas debates
overthedesign ofInternet architecture. For example, theopponents ofnetwork neutrality
argue that itrestricts theprovision ofQuality ofService (which can enable services such
as faster video streaming for customers willing to pay more) whiletheproponents argue
thatgoing against network neutrality would involveviolation ofthedesign principles of
theInternet andthat theresulting network would not onlybe architecturally incoherent
butalso less conducive to future innovation(Schewick, 2012).
However,since thecentral aim ofthisthesis is to provideastep towards afuture Internet
architecture, theanswer to thequestion ofhowthesedebates should be settledwithin the
contextofcurrent Internet is not particularly important.As mentioned earlier,designing a
clean-slatearchitecture not onlyprovides an opportunity to cater to someofourcurrent
problems, butalso aresponsibilityto anticipate orcater to problems that may arisein
future.Therefore, inthis thesis, Iaim to identify somecommon underlying conditions
thatnot only give riseto themultiple challenges forcurrent Internet (such as debates over
network neutrality,cybersecurity,theeconomic challenges for service providers, conflicts
overcensorship and privacy ontheInternet, and so on), but which may also give rise to
future, albeit different, challenges forany network deployed globally onalarge scale. In
otherwords, thetask in handisto identify andaddress theunderlying conditions that give
riseto conflicts in theirgeneral, rather than specific technical forms. Idiscuss theseissues
inChapter 5,which builds uponthediscussiononthe relation between architecture and
challenges of current Internet presented in chapter 2-4. In Chapter6,I give
recommendations forthedesign ofa future network liketheInternet, bearing in mind that
10