- what are the full set of functions needed in a metaprotocol (e.g., besides soft/hard state management, security, flow/error control,etc.)?
- what other kinds/examples of resolution are there (other than Google, DNS, ARP, BGP, OSPF)? related to this, resolution is currently in support of moving data -- are there cases where you care about the resolution only (not to move data)?
1. Money flow - how to get money to flow to providers that will inves in infrastructure and new services?
2. How should we build applications to take advantage of new services offered by the network? Can't ask (human) users, applications may not know their requirements, especially a priori. How to allow expression of policies for, e.g., choosing among paths? There are applications that know how to do this -- cf. protocols for video streaming.
3. Performance - how do we add the things we need (motivation, accountability) to the forwarding operation without destroying performance? This is a question about engineering the fast path, but may also have to do with amelioration strategies - e.g., are there ways to avoid doing it in certain cases?
Research questions: Virtualization as a Network Architecture
How do you interconnect slices
a) if there are 10 slices? 10e7?
b) if there are heterogeneous technologies and a slice-based service model?
If virtualization is fundamental, how do you create a {trustworthy, available, robust, performance-sensitive, ...} resource allocation and control framework?
Given that [ILECs] own the resources and can also build virtualized services on top of that, how can [CLECs] get a level playing field? (motivates measurement; need to think about cloud-by-cloud vs. global service models)
How are new services created?
What are the guidelines for Internet architecture design informed by economic incentives which also recognize economic viability?
What information/knowledge will be needed for decentralized network management?
How can we make sure that the architecture is flexible enough for existing and unforeseeable use cases?
Content-Centric Networking
DTNs
Latency-sensitive Networking
What are the appropriate interfaces to achieve usability? To applications? To users? To operators? What role does multiplicity play here?
Should we have trust primitives? If so, what should they be?
A Robust Architecture for Diverse Mobile Wireless Networks
Mobility key architectural concern
The architecture should assume mobility as a first class design concern.
The architecture should enable seamless opportunistic usage of diverse access technologies (3G, WiFi, WiMAX etc.)
The architecture should support intermittent connectivity
Q1: Routing under mobility
How can the routing architecture be made
robust to diverse levels of mobility?
Q2: Authenticated access
How to enable authenticated access and associated privileges across diverse wireless access networks?
Q3: In-network storage
How should content be stored and retrieved in the network?
– Support for content addressability from the routing layer
What are the implications of mobility for the persistence of in-network data?
Q4: Mobility-aware stack
What are the key components of a mobility-aware network stack?
Separating names and addresses
Hop-by-hop vs. end-to-end
Opportunistic network access
How aware of mobility does the core need to be? Can mobility be managed entirely at the edge? Do cloud services need to be mobility aware?
What is the minimal functionality?
applications design to min(capabilities)
How many service models?
Why didn’t it (= QoS, security, …) work?
Do we need a clean slate?
What makes designs successful?
The power and limitations of analogies?
Lessons Learnt
User choice
why not?
interface choice … provider choice … interface issues
need technical historian / innovation propagation expert
What can’t we do?
At home / Cisco / IETF / SIGCOMM
Application-layer attacks
Anonymity
Multiple service levels
Analogies & minimal systems
Highway systems
Bits A - B
1-to-1
along the path (see: RAO)
many-to-many (but: group membership)
What can we do with traffic diversion?
Certified priority
1. How do you interconnect slices
- if there are 10? 10e7?
- if there are heterogeneous technologies? Per slice service models
2. If virtualization is fundamental, how so you create a trustworthy/available/robust/performance-sensitive resource allocation and control framework?
3. Given that [ILECS] can own the resources and can build a virtualized network on top of that, how can [CLECs] get a level playing field?
- motivates measurement
- cloud-by-cloud vs. global