Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Fi arch design-principles-new_seeds-v0.7.4

457 views

Published on

Published in: Technology, Business
  • Be the first to comment

  • Be the first to like this

Fi arch design-principles-new_seeds-v0.7.4

  1. 1. FIArch meetingFeb. 22nd, 2012Brussels, BelgiumInternet Design PrinciplesNew SeedsIoanna PapafiliAthens University of Economics and Business
  2. 2. Outline Motivation What is a new seed Seeds  Resources Awareness  Dependability Logic  Exchange of Information  Sustain the Investment Discussion 2
  3. 3. Motivation So far we have seen existing principles that should be preserved / adapted / augmented Yet, the Internet has long evolved New challenges emerge that go beyond the networking and primitive service aspects Existing design principles may be unable to address them! 3
  4. 4. What is a new seedDefinitionA concept or a notion at the inception of a well formulateddesign principle Why seed and not principle?  Formulating principles is a complex exercise  Ongoing research in proving their value, utility and impact  Seeds may flourish while others may not ? 4
  5. 5. Resources AwarenessChallenges Resources have a physical representation and purpose Resources refer to infrastructure or services Infrastructure resources (network, storage, computation, etc.) Services: means for users to have controlled access to functionalities/information offered over the Internet  Service components, e.g., commercial services as black boxes Infrastructure is service-unaware Growth of data requires new delivery schemes that overcome limitations w.r.t. efficiency 5
  6. 6. Resources AwarenessSeed FI should include resources as first order abstractions Network should be able to manage as a whole these abstractions  Making the resource self-* explained: initial functionalities (initial abstraction) Self-description: a) Dependable b) Low-latency Initial c) ….. Abstraction: At level 1 Self-managed (Wrapper 1) Self-repair Self-… 6
  7. 7. Resources AwarenessSeed New resources should be aware of underneath resource capabilities  Making the resource self-* explained: initial functionalities (initial abstraction)  Improving and adding new functionalities/capabilities: the polymorphism concept Self-description: a) Dependable Advanced b) Low-latency Initial Service as: c) ….. Abstraction: Abstraction At level 1 At level n° Self-managed (Wrapper 1) (Wrapper n°) Self-repair Self-… 7
  8. 8. Resources AwarenessImpact Service Delivery Infrastructure optimization exploiting service behavior informationInteraction with other principles “Modularization by layering”  Enhanced cross-layer cooperation based on functionality descriptions “Loose-coupling”  Minimized interactions and non-linear effects through cross-layer awareness “Locality”  Reduced distance between processes and data 8
  9. 9. Dependability LogicChallengesLack of methods/means for reliable, accountable and verifiableprocessing and handling of inftrastucture w.r.t. services  Services not a ‘cure-all’ – not cognisant of users’ needs  Services operate on a ‘best-effort’ basis  Little or no service guarantees to the end-user  Services are modelled prior to their deployment  Infrastructure is service-unaware 9
  10. 10. Dependability LogicSeedFuture Internet must be imbued with and extend the Principles ofDependability  Availability And be further enhanced by:  Reliability Transparency Self-adapation and self-learning capability  Accountability  Verifiability  Safety  Integrity  Maintainability  Confidentiality 10
  11. 11. Dependability Logic  The definition should cover Transparency Transparency: the ability to inspect and introspect a service so that the delivered and guaranteed quality of the service agreement can be verified and observed.  24 years of work. Finalised in 2004 (IEEE)  Little consideration for the Internet of services, where end-users are service- and not product-oriented22/02/2012 “Future Internet Architecture Design Principles” 11
  12. 12. Dependability Logic Impact  Tackles the issue of lack of trust of interest to many enterprises/SMEs, service broker, users  Use of many cloud services today (IaaS, PaaS, SaaS)  Service providers may offer guarantees  Very difficult for the customer to verify these  Requires a new introspection capability*  Both in Service and Network domains * Recognized by the UK in the areas of converged Future Internet services: “Fundamental Limitations of Current Internet and the path to Future Internet”22/02/2012 “Future Internet Architecture Design Principles” 12
  13. 13. Dependability Logic Interaction with other principles  “Modularization by layering”  Application: Two or more levels of dependability management  “Loose coupling”  Application: Management differing levels of abstracted services and resources  “Polymorphism” principle  Application: Service self-adaptation based on promised guarantees; different characteristics as parameters of the polymorphism behavior22/02/2012 “Future Internet Architecture Design Principles” 13
  14. 14. Exchange of Information between communication end-pointsChallengesInternet evolved to a playground of various stakeholders  Stakeholders perform individual optimization,  Stakeholders ignore information related to others • Possibly beneficial to themselves and societyInternet is unpredictable!  Designers cannot foresee whether an outcome will be the most efficient and beneficial one  Outcomes should not be imposeTherefore:Address Information AsymmetryAllow for variation in outcome* *(proposed by D.Clark et al.) 14
  15. 15. Exchange of Information between communication end-pointsSeed Exchange of information between end-points of communication  If it serves the incentives of stakeholders / if it is beneficial  Allow stakholders to express their interests/choices overlay Constituent elements  Abstraction of information underlay • Do not expose critical information  Exposure of information ISP1  Collection of information ISP3  Assessment of information  Decision making ISP4 15
  16. 16. Exchange of Information between communication end-pointsImpact Design and deployment of more “open” systems and interfaces Anticipate users’ reactions when QoE is unsatisfactory Compatible with Information-centric networkingInteraction with other principles This is an enabler of Clark’s et al. ‘design for tussle’ principle Compatible to the end-to-end and modularization principles  Complexity mostly to reside at the edges – isolation, generality Can enable locality 16
  17. 17. Sustain the InvestmentChallenge: coopetition  result of competing antagonistic actions due to conflicting interests between parties / actors  Negative global return  Ex. DPIObjective: Internet to be designed tosustain brain and resourceinvestment toward a global positivereturn (not to sustain detrimentalconflicting interests) 17
  18. 18. Sustain the InvestmentSeed Sustain the Investment (brain, innovation, resource) toward a global positive return Internet should not be ‘designed for Tussle’ only ...but so as to lead to positive global returnThe question ... of course "how to determine homogenize user satisfaction, utility functions and individual interests across the entire Internet ?" ... and address incentives 18
  19. 19. Sustain the Investment In practice  Structure Internet design to allow various communities and peoples active involvement but without impeding its genericity, evolutivity, openness, and accessibility  Reward architectural modules and components delivering positive returns – functional module strength -progressively deprecate modules with negative return – weak modules  Many ways to model such system ...
  20. 20. ConclusionResources Awareness Include resources as first order abstractionsDependability Logic Imbue services with attributes of Dependability Extend Dependability to include the aspect of TransparencyExchange of Information Exchange of Information for more efficient and beneficial operationSustain the Investment Design/reward modules that deliver positive global return 20
  21. 21.  Many thanks to:  Francesco Torreli (Eng)  Massimo Villari (UniMe)  Dimos Kyriazis (NTUA)  Andrew Edmonds (Intel)  Dimitri Papadimitriou (ALBLB)  George D. Stamoulis (AUEB) for their help and input for this presentation! 21
  22. 22. Thank you for your attention! Questions?
  23. 23. Back-up
  24. 24. Resources AwarenessChallenges Resources have a physical representation and purpose Infrastructure resources:  Network  Storage  Computation  Any kind of resource derived from hardware Service components  E.g., commercial services as black boxes 24
  25. 25. Resource awareness: for the infrastructure (HW) and software components (SW)Infrastr. Resources SW Services FIArch seeds Growing up 25
  26. 26. Dependability Logic Brief Description  Defined by 24 years of work by IEEE  Finalised in 2004  Little consideration for the Internet of services, where end-users are service- and not product-oriented  The definition should cover Transparency Transparency: the ability to inspect and introspect a service so that the delivered and guaranteed quality of the service agreement can be verified and observed.22/02/2012 “Future Internet Architecture Design Principles” 26
  27. 27. End-points of communicationDifferent layers of different players Different players at the same layer e.g., overlay traffic management e.g., mTCP, reECN Different layers of the same player e.g., DPI, mTCP
  28. 28. Exchange of Information between communication end-pointsInter-layer overlay optimization max profit overlay networktraffic matrix traffic demand underlay network traffic management min operational cost 28
  29. 29. Exchange of Information between communication end-pointsInter-layer overlay optimization max profit overlay networktraffic matrix ‘advice’ - abstract traffic demand network information underlay network traffic management min operational cost 29
  30. 30. Exchange of Information between communication end-pointsInter-layer overlay optimization max profit overlay networktraffic matrix overlay information QoS impact ‘advice’ - abstract traffic demand network information underlay network traffic management min operational cost 30
  31. 31. Exchange of Information between communication end-pointsInter-layer Inter-player ISP2 ISP1 overlay optimization tier-2 tier-1 max profit overlay network ISP2 tier-3traffic matrix overlay information global QoS impact rendezvous ‘advice’ - abstract traffic demand network information ISP4 tier-2 underlay network transit link traffic management peering link min operational cost rendezvous 31
  32. 32. Exchange of Information between communication end-pointsInter-layer Inter-player ICN ISP2 ISP1 overlay optimization tier-2 tier-1 max profit overlay network ISP2 tier-3traffic matrix overlay information global QoS impact rendezvous ‘advice’ - abstract traffic demand network information ISP4 tier-2 underlay network transit link traffic management peering link min operational cost rendezvous 32
  33. 33. Exchange of Information between communication end-pointsInter-layer Inter-player ICN ISP2 ISP1 overlay optimization tier-2 tier-1 max profit overlay network ISP2 tier-3traffic matrix overlay information global QoS impact rendezvous ‘advice’ - abstract traffic demand network information ISP4 tier-2 underlay network transit link traffic management peering link min operational cost rendezvous 33
  34. 34. Applications of the Exchange of Information Overlay traffic management  The overlay asks underlay for advice on resource selection Multipath TCP  Congestion information is carried by flows; the end-host makes decision on how to shift load among flows Re-ECN  Congestion information is made available to any node of the network Information-centric Networking (ICN)
  35. 35. No contradiction to the “end-to-end” principle?Different possibilities for means of information exchange: Information is exchanged in packets  No impact on routing If information is carried in the fields of the protocol headers  Limited impact on routing Complexity mostly to reside at the edges

×