I plan to distribute OpenFlow switches, which is why I would like to further dis
cuss about this technology with NANOGers.
Firstly, to get a better overview of OpenFlow, I would greatly appreciate to in
vite you to a further reading of the latest presentation on the subject from NEC
entitled "Future Internet Research Using OpenFlow" [1].
I) Definition
a) Commercial
I'd note that Flow-based networking is not a new concept and is currently suppor
ted by Anagran [2], whose founder is none other than Larry Roberts [3].
b) Dynamic Switch Control Protocols
By the way, I would greatly appreciate to invite you to a further reading of th
e thesis entitled "Implementation and Evaluation of a Network Element Control Pr
otocol" [4] where different Dynamic Switch Control Protocols (Strengths, and GSM
P OpenFlow) are compared (page 17-23).
-Note: the source code is available at the web page titled "Work ETNA Package 4
II) Business Model
1) Capex
The advantage of OpenFlow is to separate the Data Plane from the Control Plane,
thus simplifying the switches (see page 15 of "Future Internet Research Using Op
enFlow "[0]) which results in a reduction of Capex (see page 23 of the presentat
ion entitled "OpenFlow / Software Defined Networks" [6]).

From a cost-perspective you can deploy a cheaper networking gear basedon "dummy

OpenFlow-only" switches, provided a controller applicationthat takes care of th
e network intelligence (e.g., installing flows,VM directory, etc.).
To illustrate this argument, I would greatly appreciate to invite you to a furth
er reading of the presentation entitled "OpenFlow / Software Defined Networks" [
6] from Clean Slate project at Stanford and the one entitled "Data Center Networ
ks Are in My Way" [7] from Cloud Provider, Amazon.
2) Opex
Moreover, intelligence is centralized through the OpenFlow controller therefore
easier, resulting in a reduction of Opex (see page 26 of the presentation entitl
ed "OpenFlow / Software Defined Networks" [6]).

From a functional perspective you have the potential gains ofre-thinking the pa

cket forwarding departing from spanning trees orVLAN segmentations (cf. SIGCOMM0
9 Portland paper).
To illustrate this second argument, I would greatly appreciate to invite you to
a further reading of the presentation entitled "An Experimenter's Guide to Open
Flow" [8] of Deutsche Telekom R&D Labs and the one entitled "OpenFlow : Operatio
nal Experiences" [9] from the GRNOC at the University of Illinois (who will also
give an OpenFlow track at NANOG50 [10])
-Note: as an example of research application, I would greatly appreciate to invi
te you to a further reading of the presentation entitled "Towards a Flow-level N
etwork Security System "[11].
III) Open Source
For those wishing to experiment with this technology, I would advise the followi
ng setup:
-HW :
Pronto 3290 [12]
-Firmware :
Pica8 XorPlus 1.1 that includes the L2/L3 management for VLAN, LACP, STP/RSTP, L
LDP, OSPF, RIP, static route, PIM-SM, VRRP, IGMP, IGMP Snooping, IPv6, Radius/Ta
cacs+ as well as OpenFlow 1.0 (available at later Oct, 2010) [13]
-OpenFlow-Controller :
Beacon: Java OpenFlow Controller [14]
Have a good week-end :slight_smile: !
I look forward to your answer,
Best Regards,

[1] [1]
[2] [2]
[3] [3]
[4] [4]
[5] [5]
[6] [6]
[7] [7]
[8] [8]
[9] [9]
[10] [10]
[11] [11]
[12] [12]
[13] [13]
[14] [14]

Guillaume FORTAINE
Tel : +33(0)631092519
Mail :

Wow, resorting to using a spoofed email address to propagate your spam,
and forget to remove your .sig

Some people just don't take a hint, do they? I know which software
package I WON'T be recommending to anyone (in fact, quite the opposite!)

Ken Matlock
Network Analyst
Exempla Healthcare
(303) 467-4671

And also, there is a nice agenda item at NANOG50;

Track: Open Flow
Nick McKeown, Stanford University; Matt Davy, Indiana University
Presentation Date: October 4, 2010, 4:30 PM - 6:00 PM

OpenFlow: An Update
OpenFlow Trials and Deployments

Thus as it is a NANOG-ish topic, I wonder why somebody needs to hide.


There is another related item planned for NANOG50:

It could be because his dumb ass got the banhammer from nanog
Mods: can you pleeeease get rid of him again?

Which is fine and all (being that it's on-topic). My main beef is that a
certain person can't take a hint. Using an 'anonymous' re-mailer to try
and get people to read nothing more than copy/paste, and then 5 billion
'references' (most of which use asinine '' references
instead of the actual document) strikes me as unprofessional at the
least, if not infantile.

I'd have thought he'd have learned the last 50 times he got smacked down
for this. But of course, I'm just an ignorant American! :slight_smile:

Ken Matlock
Network Analyst
Exempla Healthcare
(303) 467-4671

If you where less ignorant and more ignoring then nobody else would
notice it due to their killfiles...

aka 'be quiet and the trolls won't have any fun'. The mods are doing
quite a fine job already from what I heard, they can't always be on
guard though, sometimes they drink a bit too much whiskey :wink:



Geeze, do some people never take the hint?

We all would love too but dumba$$ keeps getting new domains & email addresses. I think he ate lead paint as a kid or something. He is absolutly 190% insane
Mods:: please show gilliam the door :slight_smile: