module optics with openflow

[Thiago Allisson]

Hi!

I'm working with the optics module available in ns-3-dev-pwns, hosted https://bitbucket.org/pwns/ns-3-dev-pwns/src/38a0d397386f?at=pwns-project. However need to combine open flow with wdm network. Any suggestions for how to do?

[Tommaso Pecorella]

Hi,

OpenFlow is a system to decouple the control plane and the data forwarding plane in Switches.

The optic module implements an optical switch network (along with the optical devices, like ONUs). This, of course, if I do remember right, since the module is an "external" module.

Mind if I ask you why *we* should have an idea on what to do with respect to the research topic that *you* are talking about ?

I see a common pattern recently. People (like you) that comes with perfectly legit questions that seems to come out from an evil / idiot tutor that just gave the poor student a one-line topic to elaborate / discuss for their graduation thesis.

The problem is: we can not give you the ideas that you lack.

Despite the emphatic link that I can have with students (or researchers) left alone with nothing more than a vague (and possibly wrong) research topic, made by combining random buzzwords that can be mixed just like salt water and crude oil (try by yourself)... despite this empathy, I have to say that no, we can't give you [good] ideas.

In a world where ideas are research, papers and, ultimately, money, giving ideas for free is not a good idea.

As a consequence, sorry, but even if a reader would have good research ideas for "optics + SDN" mix, I don't really think he/she would simply give you them. It's not a matter of being nice, it's a matter of being logical.

Sorry if it seems harsh, I'm just realistic.

Of course, programming issues and specific development questions are a different thing.

Good luck,

T.

PS: the ns-3 OpenFlow module is not really dependant on CSMA. Judging from the code (I didn't try this out tho) any interface with a 48-bit MAC should be ok, even an optical one.

The problem may be to use the OpenFlow concepts in an optical switch, but I guess that even THINKING to an OpenFlow approach for an optical switch is a sort of heresy. 

Think about it: a packet arrives to the optical interface, the switch doesn't know what to do and ask the controller, the controller decides and push to the switch a configuration. That configuration is used for the packet and all the similar ones following the first. How big it should be the buffer in the switch? Considering an optical 200 Tbps interface, and a fast controller (let's say a few milliseconds...) LOL, srsly ?

[Thiago Allisson]

Thanks! 

Sorry, I expressed myself wrongly. 

When I said if anyone had a suggestion, was to know whether anyone had come across something similar

[Tommaso Pecorella]

Hi Thiago,

I haven't seen similar discussions on the group so far. At least not involving optical nets.

Cheers,

T.

[Vedran Miletić]

Hi Thiago, Tommaso,

I completely second what Tommaso wrote initially.

Moving to "does the optical network model support/have an usable OpenFlow inteface", the answer is "no, not yet but you are welcome to contribute patches".

Regards,
Vedran

--

Vedran Miletić, Research and Teaching Assistant
University of Rijeka Department of Informatics / Faculty of Engineering
http://vedranmileti.ch/ | vmil...@inf.uniri.hr

[Thiago Allisson]

Thanks! All right? My dear, I exchanged some ideas with Tommaso and was looking for materials relating to switching circuits with openflow. I found a thesis entitled, A UNIFIED CONTROL ARCHITECTURE FOR PACKET AND CIRCUIT

NETWORK CONVERGENCE. I'm reading it to proceed with the implementation in the appendices he presents the API. I am also based in me OpenFlow specification. Anything I contact you to report progress.

I thank you for your attention and collaboration.