[Ns-developers] GSOC 2012

V.Sindhuja intutivestriker88 at gmail.com
Wed Mar 28 10:28:10 PDT 2012


Hi,

I have gone ahead and submitted a proposal for the Network address and port
translation models. I am eagrly looking forward to the valueble feedback
from all of you.

-- 
Thanks and Regards,
Sindhuja Venkatesh
Graduate Student, Department of CSE,
University at Buffalo, SUNY.



On Sun, Mar 25, 2012 at 5:34 PM, Lalith Suresh <suresh.lalith at gmail.com>wrote:

> Hi Sindhuja,
>
> On Sun, Mar 25, 2012 at 5:29 PM, V.Sindhuja <intutivestriker88 at gmail.com>
> wrote:
> > Hi,
> >
> > Thank you so much for the response. I was going through the list of
> > criterion to select a project and came across
> > -Impact and relevance to future users of ns-3?
> > - Availability of a mentor for the suggested project.
>
> If you're picking something from the ideas page, the above two
> shouldn't be a problem as long as you have developed a strong
> proposal.
>
> > These are not things that are in my purview and I dont want to lose out
> on
> > being able to do a project based on these. Considering the NAT and
> Netfilter
> > part has been already worked on and that in IPV6 we need to add new
> > features. I am unable to chose based on these. Do help me out ! I'd love
> to
> > chose one and give it my 100% or choose all three and give a 300% ! :)
> >
>
> So really, pick one that interests you the most, and start working on
> a proposal. See below for some quick pointers on the topics you
> suggested:
>
>
> >>
> >> On Fri, Mar 23, 2012 at 9:56 PM, V.Sindhuja <
> intutivestriker88 at gmail.com>
> >> wrote:
>
> [snip]
>
> >> >
> >> >
> >> > 1) *ns-3 Firewall model*
> >> >
> >> > After having gone through the code base that exists for the initial
> work
> >> > w.r.t to Net filter is see that there exists a netfilter-based
> >> > infrastructure to filter the packets. I would like to evaluate if the
> >> > following are the additions to be made:
> >> >
> >> >
> >> >   -       Interface for adding the IPV4 rules instead of having to
> hard
> >> >   code.
> >> >   -       IPV6 support module for the filtering.
> >> >   -       Any additional firewall functionalities e.g. port knocking,
> >> >   IPSec implementation.
> >> >
> >> >  It would be great if I could get an idea on what would be focusing on
> >> > primarily so that I can focus my proposal on that.
> >> >
> >> >
> >> >
> >> > 2) *Network Address and Port Translation (NAT) models*
> >> >
> >> > Nat is an integral part of the Firewall model as well. I also see that
> >> > as
> >> > part of GSOC 2009 project by Qasim Javed has worked on both the
> modules.
> >> >
> >> > Please let me know if the following would be the working direction of
> >> > this
> >> > project:
> >> >
> >> >
> >> >   - Addition of a Nat Traversal and Nat Exemption interfaces to
> support
> >> >   IPSec.
> >> >   - Multiple NAT Types implementation.
> >> >
> >> > I see that the existing model does handle the pack processing with
> hooks
> >> > and the nat priority on them. However I would like to know the primary
> >> > focus would be on this as well in order to write my proposal.
> >> >
>
> There has been some discussion on the above two on the mailing list
> over the last few days. Have a look.
>
> I'd say you're on the right track, but in any case, I've cc-ed Adrian
> to pitch in comments since he's the concerned mentor.
>
> >> >
> >> >
> >> > 3) *IPv6 stack validation and improvements*
> >> >
> >> >
> >> >   - The features that we implement are they specific to the list
> given?
> >> >   - What is the recommended size of features to be implemented? This
> is
> >> >   taking into account the depth of implementation and the time frame.
> >> > For
> >> >   example implementation of IPSec would take long time in itself
> >> > considering
> >> >   the several types we may want to add in VPN models.
> >> >
>
> Again, you're on the right track. As the proposal says, the list of
> features provided isn't exhaustive. But I think it would be reasonable
> to pick say, one large/hard or two medium ones out of the proposed
> features and get some polished code merged by the end of the summer
> rather than half implementing too many features.
>
> Out of the list provided, I'd say path MTU discovery is super
> important. So look at src/internet/model/ipv6-l3-protocol.cc, and see
> how you can fit RFC 1981 into it. And get a draft proposal ready soon
> so that you can start iterating on it with the mentors.
>
> Tommaso and Andrea should be able to guide you better on the
> scope/size of the features and what you should focus on.
>
>
>
> --
> Lalith Suresh
> www.lalith.in
>



More information about the Ns-developers mailing list