First, the sigcomm award is given to Larry Peterson, a professor in Princeton, for ground breaking advances in how networking and distributed systems research is conducted. Larry Peterson is well-known for the following things:

1. Co-author (with Bruce Davie) of a famous networking textbook "Computer Networks: A System Approach".

2. Established PlanetLab

3. He chaired planning efforts for the NSF's GENI intiative.

4. Editor of ACM TC, ACM/IEEE TON, ...

His keynote is titled "Zen and the Art of Network Architecture". I could not find the slides. The only related information i can find is from the liveblog of sigcomm 13

1. The middle way: Involves balancing requirements, not about optimising any one dimension.

2. Path to Enlightenment:

    Inspiration - Some new idea

    Analysis - Simulation, back of the envelope

    Controlled Lab Experiments - Implementation reality

    Deployment Studies - Traffic and user reality

    Pilot Demonstrations - Customer reality

    Commercial Adoption - Market reality

    Change the market

3.  PlanetLabs & CoBlitz - 12 year time frame. CoBlitz is a CDN system

     Changet the market: Operators now all have their own CDN systems.

4.  The path to enlightenment:

     See reality clearly - Assumptions hide the truth.

     Users reveal the truth faster than any other way.

5.   Architectures tell engineers what they can not do, but engineers don't read achitecture documents.

6.   Lessons on architectures:

     Part analysis, part intuition - whole is greater than sum of the parts

     Unify abstractions  - duality is an opportunity. (对偶？)

7.   Putting Lessons to Action

      SDN vs NFV (Network function virtualisation) vs Distributed Applications

      Lines are quite blued. Distinctions without a difference

          Is a proxy an application or controller?

          Is a scalable controller that uses NoSQL an app or a controller

          Is a firewall in the data plane or in the control plane?

      Topology - We all have a network graph in our heads.

           Network people like interesting topologies, but applications like a bit switch. Everything to everything.

           Basic model: Application is a function that sits on top of the network switch

           There are topology optimisations

                cut-through - don't bother sending data up to the application

                Inline - put the function in the network and don't bother sending to the "application"

            Really - Cut-through = SDN, inline = NFV

       Model all network functions as "scalable services"

       Use SDN to bootstrap a virtualization layer - supports cut-through optimisation

       NFV - tool to decide where to put functions.

8.  XaaS - Everything-as-a-Service

       Service as a unifying abstraction

       Unifies across resources (Compute, Network, Storage)

       Unifies across the network (DC, Access, WAN)

       Unifies across service levels (IaaS, PaaS, SaaS)

9.  Questions:

       Q: What about hardware? A: This model applies as well. Keep it simple, keep it fast and keep the interfaces static.

       Q: PlanetLabs: Haven’t we been doing OpenCloud all along: A: Network virtualisation is coming up to speed with server virtualisation.

       Q: What about publishing: A: Operational experience should be a SIGCOMM category.

       Q: What about incentives. Aren’t there conflicting requirements encryption vs storage. A: Hard

[Ref Link: http://www.syslog.cl.cam.ac.uk/2013/08/13/liveblog-from-sigcomm13-day-1/] 

Second, the best paper as well the best student paper is "Ambient Backscatter: Wireless Communication Out of Thin Air". The main idea is using ambient RF as the power source of communication. Very cool and green idea. If it can be adopted by industry, much energy can be saved!!!