Intro

https://www.youtube.com/watch?v=JLAfLWE76fE

Foundation of global internet

1972: open network

• Key arch

• Interface: plug and play

• Internet: over-arching notion of how to federate different networks together

  • A logical framework

  • Layering: learn from early network

    • Expose: someone can interface different parts of the systems

    • Implementation strategy: modularization

    • Not a requirement

    • 1970s: setting layer specification, but nobody implements, just a ref model --> need more tightly-integrated implementation

  • Gateway linking?

    • I.e. routers

  • Packet: have the system self-configured so it knows what pieces were connected in where

    • Routing protocols

  • Assumptions on the Internet

    • Binding of addresses etc. didn't have a lot of mobility back then

    • These assumptions might not hold today (i.e. new applications and new needs)

    • 1983s: community more evolves --> nobody knows enough on the history to do a arch change

  • Why don't you make a more secure system?

    • Technical challenge, will also be social

    • Edge of the net: where the IP layer ends and the host computer starts, a bunch of interesting protocol happen inside in an end-to-end basis

      • End-to-end notion: much of the functionalities are in the edges; no permissions from IP part to try things

  • Evolve under different contexts?

    • One answer

      • Interfaces that define different protocol layers: stability

        • I.e. optical fiber, switching tech

      • Open place to innovate (?)

        • I.e. P2P don't have to standardize by anybody

    • The other

      • Research community involves in the back

      • Now: internet as a net

        • Business model, what changes to do with customers

        • Originally bottom-up scenario

        • Today: lack of changing things in a big architectural way

          • But big thing is pretty small

          • What would it take to make a big architectural way?

            • Happen in incremental changes

            • Or creation of something completely independent

      • New technology: new? or replacements?

        • Grow up in existence of the switching

        • New techs will co-exist, and run on-top of the internet for a while

          • Nothing is instantaneously happening

      • Virtual private network

        • Interact through virtual circuit

  • Applications: sub-optimal

    • E.x. voice network?

    • One hand: made to work with lots of different applications

    • But: remove motivations --> cross-boundaries to make something different

    • Harder: mobility, identity management and auth strategies; real change in Internet is pervasive (i.e. something beyond the edge, software, hardware, distributed sys, application in some sorts --> whole new set of challenges)

    • Assumption

      • Wireless: most of the time disconnected

      • But back then: assumption is that mostly connected

      • Uniqueness is valuable!

        • i.e. identity

        • But lack of commonality

        • Network partitioning: move away from the current architecture; maybe the early binding is too tight