Search Mailing List Archives


Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[mininet-discuss] Running a Click router in Mininet

Shrutarshi Basu basus at cs.cornell.edu
Fri Oct 14 08:19:04 PDT 2011


I'm starting to work on building a Mininet wrapper for Click. Is there a
specific branch I should be working on if this is something I would like to
eventually push back upstream? I'm planning on adding my class to the the
node.py file along with the other switch types.
Thanks,
Basu

On Fri, Oct 7, 2011 at 11:53 AM, Bob Lantz <rlantz at cs.stanford.edu> wrote:

> One final thought though - I think implementing something in Click is
> harder (and possibly less useful if you don't have Click-based hardware)
> than using OpenFlow, but easier than changing Open vSwitch!
>
> On Oct 7, 2011, at 8:36 AM, Bob Lantz wrote:
>
> [Ah, make that NetFPGA/WARP/network processors/bee3/any of the other
> various hardware prototyping platforms that are available and that I seem to
> be forgetting off the top of my head but that might be well-suited for a
> click-based implementation path.]
>
> On Oct 7, 2011, at 8:18 AM, Bob Lantz wrote:
>
> Well, the mailing list is about Mininet, but:
>
> 1. There's no reason you can't do software or hardware DPI with OpenFlow
> (e.g. either external processing or OF extensions.)
>
> 2. I think the only reason to use Click is if it gives you a faster path to
> hardware implementation (e.g. via NetFPGA or Bee2 or whatever.) I still
> advocate controlling it using OpenFlow.
>
> 3. DPI is of limited value in the internet (dominated by encrypted video
> streams), in the cloud (encrypted traffic and managed apps), and in the
> datacenter (managed apps.)
>
> (Hmm, thinking back to my experience working for a cellular operator I'm
> not sure it's of much use on cellular networks either.)
>
> Anyway, good luck, and we'll look forward to the ClickSwitch class!
>
> -Bob
>
> On Oct 7, 2011, at 6:49 AM, Shrutarshi Basu wrote:
>
>  Hi,
> Thanks for the quick reply, I'll see what I can do. The reason I want to
> run Click is because I want to do some work involving heterogenous where
> some switches are OpenFlow and others have some extra capabilities that are
> outside the OpenFlow Spec (like DPI for example). I'll let you know how it
> goes (and maybe send in a patch).
> Thanks,
> Basu
>
> On Thursday, October 6, 2011 at 10:40 PM, Bob Lantz wrote:
>
> p.s. A major advantage of OpenFlow is that once you have a design working
> in Mininet, it's easy to move to a hardware OpenFlow switch.
>
> On Oct 6, 2011, at 7:38 PM, Bob Lantz wrote:
>
> I'd recommend making a subclass of mininet.node.Switch to start/connect to
> your click router, the same way we start Open vSwitch.
> If you get it working, please submit the patch to us!
>
> I do have one question though: Is there some reason why you can't do what
> you are trying to do with OpenFlow?
>
> If you can do whatever it is you are trying to do using OpenFlow, then you
> can just use Mininet as-is.
>
> -Bob
>
> On Oct 6, 2011, at 7:15 PM, Shrutarshi Basu wrote:
>
> Hi,
> I'm considering some experiments involving running a custom Click router as
> part of a larger network of hosts and switches. Has anyone successfully
> virtualized a Click router in Mininet? Are there any gotchas or pitfalls I
> should anticipate?
> Thanks,
> Basu
> _______________________________________________
> mininet-discuss mailing list
> mininet-discuss at lists.stanford.edu
> https://mailman.stanford.edu/mailman/listinfo/mininet-discuss
>
>
>
>
>
>


-- 
--
Shrutarshi Basu
Basus.me
The ByteBaker <http://bytebaker.com> -- because Computer Science is not
about computers
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.stanford.edu/mailman/private/mininet-discuss/attachments/20111014/e3f24f90/attachment.html>


More information about the mininet-discuss mailing list