[Ns-developers] customizable ns2 ? (ns-3 and future development)
Adrian Sai-wah Tam
adrian at ieaa.org
Sat Jun 3 02:09:09 PDT 2006
Agree. Actually, as the project grows, I see NS2 becomes more and more
difficult for a beginner to study. I can see that if there is a student
who has a very in-depth knowledge on computer networks and C++
programming, he would also found it not easy to modify NS2 to implement
something he wants.
I know that it is not good to modify NS2 now, but for NS3, I believe we
should re-design the structure to make it more scalable. First thing I
believe we should do is to really implement a full-featured protocol
stack so that we can interface the simulator with the real network.
Second thing is, as Batov said, make it "customizable", so that we can
put in or move away some "modules" easily.
I quite appreciate the Linux kernel's module structure, which allows you
to add/remove some components in compile time *without modifying the
code*. However, do anyone has a plan for NS3 yet? Any architectural
decision made?
Batov Igor Sergeevich wrote:
> Is there any plans to make ns2 more customizable?
> By "customizable" I imply means to "extract" from huge ns-allinone sources
> rather small and clean simulator that contains only code for models necessary
> for specific purposes.
>
> The idea is to make unifier that combine different models into one simulator.
>
> For example one may want only models of Static routing, Agent/FullTCP, DropTail and Webcache in its simulator. And don't want to consern all stuff related to Wireless networks, MPLS and so on...
>
> As number of contributed models will grow (and even number of models for the same object, which differ in abstraction (granularity) of simulation) I think it is necessary to have such "extractor" that will provide code that is simple to analyse and verify.
> This will also improve effectivness.
>
> What is the best way to implement such feature in your opinion?
>
> Best regards,
> Igor Batov
>
>
More information about the Ns-developers
mailing list