Transcript Future Internet

Future Research Directions
Jennifer Rexford
Advanced Computer Networks
http://www.cs.princeton.edu/courses/archive/fall10/cos561/
Tuesdays/Thursdays 1:30pm-2:50pm
Internet Design Based on Common Goals
• Original design of the Internet
– “Hook all the computers in the world together so
that as yet unknown applications could be
invented to run there.”
• Today’s reality
– “The Internet is not a single happy family of
people dedicated to universal packet carriage.
There is contention among the players.”
Stakeholders With Different Goals
•
•
•
•
•
•
Users running applications
Commercial providers making money
Governments enforcing laws
Intellectual property rights holders
Malicious parties who want to do harm
…
Contention (Tussle) Amongst the Parties
•
•
•
•
•
•
•
Single IP address vs. use of NATs
Property rights vs. P2P file sharing
Government wiretapping vs. encryption
Firewalls vs. tunneling, rerouting, port tricks
Robust, efficient routes vs. ISP competition
End-host congestion control vs. selfish users
…
Design for Tussle
• Tussle in the Internet takes place at run time
– Not primarily at design time (i.e., IETF)
• Yet the design affects how tussle plays out
– What each component is capable of doing
– The boundaries between different components
• Designing for tussle
– Design for choice, for variation in outcome
• Open interfaces; separation of policy and mechanism
– Modularize the system along tussle boundaries
• Bad: DNS names to name hosts and express trademark
• Good: ToS bits separate application from service quality
Discussion
Research Challenges
Research Challenges
• Improve system properties of the Internet
– Reliability
– Security
– Managability
– Scalability
– Performance
• Provide new features in the Internet
– Mobility and disconnected operation
– Interactive applications
– Energy efficiency
– Innovation inside the network
Tension Between Goals
• Mobility vs. scalability
– Location-independent addressing…
– … vs. hierarchical addressing tied to routing
• Reliability vs. affordability
– Redundancy and avoiding shared risks…
– … vs. co-location of nodes and links
• Security vs. innovation
– Limiting the power of the end system
– … vs. programmability for new capabilities
• Security vs. privacy
– Self-certifying identifiers and attribution
– … vs. anonymous communication
Areas That Interest Me
• Network management
– Protocols and monitoring for ease of management
• Programmability inside the network
– Enabling (rapid) innovation and customization
• Network virtualization
– Parallel virtual networks and new management apps
• Incremental deployability
– Backwards compatibility and economic incentives
• Rigorous protocol design and analysis
– Optimization theory as a way to “derive” protocols
• Energy efficiency
– Green networks and reducing energy at end hosts
Areas That Interest You?
• What topics strike you as most important?
• What research approaches seem most
appropriate?
• What are your thoughts on the collection of
papers we read and discussed?
Conclusions
• Tons of scope for interesting research
– Intellectually fascinating in its own right
– … and many connections to other disciplines
– Practically relevant, with chance for real impact
• Written report for projects
– Due “Dean’s Date” (Tue Jan 11)
– https://dropbox.cs.princeton.edu/COS561_F2010/Final_Project
• Next and final class
– Course project presentations (15 min each)
– Friday January 14 12:30pm-3:30pm
– Lunch provided!