Future Internet Evolution @ IETF76

From $1

Wednesday 11th November, 2009, 1145-1430 hrs, Room Castleview 1, ANA Crowne Plaza Hotel, Hiroshima, Japan

Presentations, brief notes

Agenda Mat Ford (ISOC)

Mat introduced the meeting and briefly discussed the agenda. Feedback on the value of this meeting and ideas for presentations and topics for discussion at future meetings are welcome. Attendees introduced themselves and took the opportunity to briefly mention recent research results.

AKARI - Overview Hiroaki Harai (NICT)

Harai-san provided an overview of the scope of the AKARI project.

AKARI - ID/Locator Split Ved P. Kafle (NICT)

There was interest in a longer discussion on 'lessons learned from IETF activities' in this area. Similarly the question of how one evaluates the extent to which such mechanisms really 'help' with scalability, mobility, etc. could use more elaboration.

AKARI - Network Virtualisation Aki Nakao (Tokyo Uni)

Discussion focussed on the motivations for this work. It was observed that security (segregating customer traffic) is one common motivation. This work is being done in collaboration with GENI. There was a lengthy discussion about the motivation for using a virtual machine to host each network slice, and the shortcomings of such an approach. It was also unclear whether support for multiple network architectures was really something application developers were asking for.

ATLAS Internet Observatory Danny McPherson (Arbor Networks)

Lots of good data giving rise to interesting questions such as, 'Is bandwidth growth a threat to Internet stability?' What potential is there for reducing the duplicated output of the 'hyper giants'? As networks become tuned to specific content and usage patterns, rapid change could cause disruption. Danny can supply data if people have specific questions they want answered - there's about six different sources for this work.

OpenFlow Switching Atsushi Iwata (NEC)

If you need hardware support on the openflow switch to do e.g. IP forwarding, encryption, etc., then do you end up with any benefits after adding all the hardware that you stripped out of a router back in?

[Atsushi Iwata] Openflow only moves routing and signaling software from router 
box to outside box, without changing the hardware side. Therefore, this 
statement is a bit wrong. We do not lose any hardware functionality using 

To make this scale, flows have to be aggregated towards the network core. How much flexibility is there really to try out different routing paradigms if the functionality required isn't present on the switch?

[Atsushi Iwata] This statement is a bit different. Basically, the same routing 
algorithm can be utilized using Openflow. We don't lose any functionality 
using the same hardware. Basically if we use Openflow, we can increase 
the flexibility and routing optimization. The point is that we don't lose 
anything using OpenFlow. Because the software installment point is 
different from the current router platform. This allows us to change the 
software in a standardized API.

Viewpoint seems to be that starting with a limited functionality device with the potential to evolve is better than starting with something more sophisticated to begin with. Useful for separating production traffic from experimental traffic and control plane from data plane - could help with scalability for testbeds.

WIDE Project's Future Internet Hiroshi Esaki (Tokyo Uni)

WIDE project has been going for over 20 years - how much longer? Don't know, but they continue to do their best!

LIPSIN: Line Speed Publish/Subscribe Internetworking Christian Vogt (Ericsson Research)

Discussion focussed on the need for scalability measurements to prove that this truthfully addresses scalability concerns. It isn't agreed that there is any fundamental scalability limitation in the current architecture to motivate this sort of approach. At a 10,000 foot level this is the same model as IP multicast, but the forwarding model is different. Forwarding by context is hard - mapping to a numerical value should help to reduce overhead and improve performance. Some interest was expressed in exploring the applicability of this to improving the performance of DTN.


Name Affiliation
 Mat Ford
 Ryuji Wakikawa
 Hiroshi Esaki
 Tokyo Uni, WIDE
 Joe Touch
 Leslie Daigle
 Danny McPherson
 Arbor Networks
 Martin Roehricht
 Karlsruhe Uni
 Stephen Farrell
 Trinity College Dublin
 Karen O'Donoghue
 Lucy Lynch
 Lixia Zhang
 Matthias Waehlisch
 Free University Berlin
 Thomas Schmidt
 Hamburg Uni
 John Schnizlein
 Jukka Manner
 Helsinki University of Technology
 Sebastien Decugis
 Lars Eggert
 Phil Roberts
 Kevin Fall
 UC Berkeley
 Hideaki Imaizumi
 Tokyo Uni
 Sangjin Jeong
 Dae Young Kim
 Youngseok Lee
 Dimitri Papadimitriou
 Alcatel Lucent
 Atsushi Iwata
 Kenji Fujikawa
 Fumio Teraoka
 Keio Uni
 Hiroaki Harai
 Ved P. Kafle
 Christian Vogt
 Ericsson Research


FileSizeDateAttached by 
Agenda - Mat Ford
36.03 kB10:07, 18 Nov 2009Ford@isoc.orgActions
AKARI Network Virtualisation - Aki Nakao
1972.98 kB14:41, 17 Nov 2009Ford@isoc.orgActions
Openflow Switching - Atsushi Iwata
3.09 MB15:11, 17 Nov 2009Ford@isoc.orgActions
 Line-Speed Publish-Subscribe Inter-Networking (Christian Vogt).pdf
LIPSIN - Christian Vogt (for Petri Jokela)
179.55 kB10:08, 19 Nov 2009Ford@isoc.orgActions
 WIDE Project_IETF_Nov10_2009.pdf
WIDE Project Research - Hiroshi Esaki
6.46 MB09:51, 18 Nov 2009Ford@isoc.orgActions
AKARI Overview - Hiroaki Harai
1012.83 kB14:41, 17 Nov 2009Ford@isoc.orgActions
ATLAS Internet Observatory - Danny McPherson
4.76 MB15:08, 17 Nov 2009Ford@isoc.orgActions
AKARI ID/Locator Split - Ved Kafle
284.61 kB14:41, 17 Nov 2009Ford@isoc.orgActions
Images (0)
Comments (0)
You must login to post a comment.