Kostas Pentikousis

Home :: Short Bio :: Contact :: Publications :: Talks :: Tutorials :: Software :: Supervision

Traffic Monitoring and Analysis for Future Internet Architectures

Talk presented at the COST IC0703 TMA Management Committee Meeting, Pythagoreion, Samos Island, Greece, 22 September 2008.

Slides—pdf

Tags—TMA, Future Internet, 4WARD

Traffic Monitoring and Analysis: A Post IP-deployment Story

Fact: The DARPA protocols were tested in overprovisioned "testbeds" (read: minicomputers) connected with overprovisioned operational networks used in a different way

You have Telnet and FTP. What else do you need? -Alex McKenzie, 1975 (a quoted in J. Day, Patterns in Network Architecture: A Return to Fundamentals

Fact: Essential monitoring and management tools and protocols came at least 10 years after the network was put into actual use

• traceroute (87?)
• ping (90?)
• libpcap (94?)
• DAG (94?)
• SNMP (1988?)

Fact: Modeling is still an unfinished business

Fact: The Internet is based on a single paradigm (packet-stream communication)

TCP/IP is Dead. Long Live FI!

There is a significant effort to move beyond TCP/IP and introduce new paradigms and Future Internet architectures (FI)

• Several EU- and nationally-funded projects are looking at different paradigms, including 4WARD and ICT SHOK; ditto in North America and the Far East
• Several infrastructure/testbed FI projects worldwide
• Some to be used for testing proposals from other projects
• Don't want to mention any names, but I bet they are all overprovisioned :)

At this stage, feasibility matters for FI proposals

Traffic Monitoring and Analysis (TMA) issues may be overlooked; evaluation using "simple" networks or simulation

Assuming that all the FI proposals are deployed by 2015, shall we repeat all TMA work in 2025, this time for FI?

Beyond Packet-stream Communication

New paradigms emerge

• Bio-inspired
• Information-centric/dissemination networks
• Publish/subscribe
• Locator/identifier split
• Network coding
• Objects, clouds, grids

New testbed architectures

What exactly "qualifies" as "FI"?

Open Questions

Status check: Which projects consider TMA already, which do not, and why? Does it really matter at this stage?

There are years of TMA experience in IP. What are the fundamental principles that we can take into the new emerging paradigms?

We still cannot model everything. How can we improve in FI?

What TMA mistakes should not be repeated?

What is our wish list for features in FI architectures?

Can we provide generic guidelines? Or, does each emerging paradigm need its own TMA-specific R∓D?

TMA as component/theme in FI or ad hoc tool collection?

Which assumptions brake?

Multiaccess prevails: Is the tap in the middle obsolete?

Expression of Interest for a SIG within COST IC0703

People from Fraunhofer FOCUS, LIP6, NEC, NTUA, Politecnico di Torino, TSSG, U Geneva, U Lusofona, U Patras, VTT, and WUT, among others, already expressed their interest to join such a SIG

Direct liaison with

• EU-funded projects: 4WARD, BIONETS, E3, EFIPSANS, INTERSECTION, ANA, EuroNF, ...
• nationally-funded projects in Finland, Poland, Switzerland, ...
• Infrastructures: GEANT, FEDERICA, onelab2, Panlab II, VITAL++, ...

Next Steps

Start drafting a white paper; NDAs need to be respected

First draft released to rest of COST TMA by December 2008: Scope, status check, main points covered

Final version presented at next TMA meeting (Spring 2009)

TOC, contributors, topics of interest: work in progress

Initiate a mailing list

Home :: Short Bio :: Contact :: Publications :: Talks :: Tutorials :: Software :: Supervision