Post on 27-Dec-2015
Edge->Core vs. Core->EdgeHow will ICN make it into the World?
Computer Laboratory
A Core-driven Story
• Two aspects potentially interesting for (core) ISPs
• Caching transit traffic
• Multicast and policy support at inter-domain level
• Problem
• Caching is not compliant with the incentive to incur traffic charges
• Same goes with multicast
-> requires changes to the incentive models for the core ISPs
-> could be driven by green, efficiency and policy aspects
But altimately: why killing my cash cow?
An Edge-driven Story
• Information dissemination is at the heart of many industries
• Retail, news, finance, content, …
• But also mobile and access ISPs (accommodating local access of information)
• Efficient delivery of what is important (information)
• Caching & multicast support
• Support of edge networks (such as wireless)
• Path computation important, e.g., for accommodating cross-value chain scenarios!
How Is It Likely to Happen?
Gov.Retail
HealthContent
Proprietary Silo Solutions
Tiered inter-connection
Core Internet
Driven by needfor internal inter-connection & cross-value chaincollaboration
Growing from edge to core
Akin to the growth of P2P
Driven by needto support policy-compliant inter-connection of information
Growing from core to edge
Questions ownrevenue!
4
Native vs. Overlay Deployment
• Both models need support!
• Overlay possibly more important for inter-domain & international deployments
• Native can be done in intra-domain solutions, e.g., service solutions for customers
• Connection to SDN and virtualisation techniques important!
• Focus on efficient intra-domain fast & slow path important!
• Get to things fast and efficient!
-> PSIRP D4.6 studied native vs overlay deployment!
What (Edge) Industries Could Push ICN?
• Mobile/Access ISPs
• Dealing with content (caching and multicast proposition)
• Broadcasters and content producers
• Highly personalised and interactive content (and increasingly over mobile)
• Finance and news
• Information at your fingertip
• Smart cities
• Internet of Things at scale
What is Needed to Make it Successful?
• Treat ICN as a new entrant compared to the incumbent IP
• Focus on secondary features: what is it that the highly content-focused IP can’t do well?
• Possible avenues:
• Interactivity? Policy compliance?
• Large degree of mashing up relevant but unhosted content?
• Sheer scale or efficiency?
• Most importantly: the new entrant ICN needs the commercial counterpart
• Candidates: an incumbent renewing itself OR the new Cisco