Routing Protocol and Table building Confusion

25 views
Skip to first unread message

Ubaid ur Rahman

unread,
Nov 26, 2015, 4:38:48 PM11/26/15
to ns-3-users
Hello again!

My problem outline: I am still confused in creating a routing protocol for FAT Tree. I read some question regarding the addressing scheme, which I agree with you ( T. ) a 100% has issues. I have restructured the addressing scheme with a little adjustment. I have depicted Pod 0 in the file attached, and I am trying to build the tables shown in that file, and my question are also typed. How should I approach this problem.

P.S How would I decide the current interface is local loop?
Routing Problem.pdf

Tommaso Pecorella

unread,
Nov 26, 2015, 6:25:46 PM11/26/15
to ns-3-users
Hi,

I'll reply here and not in the PDF, hope you don't mind.

First and foremost, there's an error in the Routing tables you have shown. it's impossible to have:
- 0.0.0.2/8 to port 3
- 0.0.0.6/8 to port 4
because the two networks are the same (0.x.x.x). moreover, it's not wise to use a "0" as the first byte. It's reserved if I remember right.

And now the questions.
A1: In RipNg the loop checks if the interface is not the loopback *and* if there is a non-link-local address. In IPv4 there's no equivalent concept for link-local, at most you'll have the non-routable address space. We never enforced the use of global Vs non-routable addresses in ns-3, but we don't guarantee that this will not change in the future (i.e., we could add a function to drop packets in routers if they have a non-routable address).
Anyway, 127.0.0.1 is "global" (not really, but we don't really care) and loopback. Tbh I don't even know if there's a function to check if an address is global in IPv4.

A2: yes, you need to create your own packet format. Feel free to borrow an existing one if it fits your needs.

A3: Don't feel dumb for not thinking to it.... CombineMask (!mask) - i.e., use the reverse bit network mask (255.255.255.0 becomes 0.0.0.255).

Cheers,

T.

Ubaid ur Rahman

unread,
Nov 27, 2015, 3:00:36 AM11/27/15
to ns-3-users
Hello,

Thanks a lot! It is clear now... And you are right 0.0.0.x/8 is not valid, I was following that paper's convention. The algorithm is supposed to check the suffix of destination address, for which such entries were given as an example. But now that I know how can I extract and check the suffix -- amazing I didn't thought of this, we don't see the simple solutions which are the right ones!! (Ah! too many things in mind :D)
Reply all
Reply to author
Forward
0 new messages