[brite-users] edge connection mechanism in top-down approach

Anukool Lakhina anukool at cs.bu.edu
Fri Sep 12 14:07:34 EDT 2003


Vladimir,

Sure, sounds like an idea with the caveat that I don't know if this is
representative.  

The edge connection methods provided in BRITE are skeletal - in the sense
that they are provided to be built on.  Maybe you could build this
functionality and contribute it to BRITE ;-)

Cheers and Good luck,
Anukool

On Fri, 12 Sep 2003, Vladimir Blagojevic wrote:

> Anukool,
> 
> I agree that smallest k-degree is a form of preferential connectivity.
> However, isn't it hard for the users to choose an appropriate k? How about
> choosing nodes randomly out of certain percentage of the most connected
> nodes?
> 
> Cheers,
> Vladimir
> 
> On Thu, 11 Sep 2003, Anukool Lakhina wrote:
> 
> > Vladimir,
> >
> > The edge connection methods at [1] are borrowed from GT-ITM.
> > A form of preferential connectivity arises from Smallest k-degree method.
> > By setting k appropriately (say to the largest degree).
> >
> > The question of what is a representative edge connection method is to the
> > best of my knowledge unknown.  There are many issues involved when two
> > ASes connect and its not immediately clear why the largest degree
> > routers in two ASes should have a link between them.  Public peering at
> > exchange points make things more complicated.
> _______________________________________________
> brite-users mailing list
> brite-users at cs.bu.edu
> http://www.cs.bu.edu/mailman/listinfo/brite-users
> 



More information about the brite-users mailing list