[brite-users] Choosing topology generation model

Anukool Lakhina anukool@cs.bu.edu
Tue Aug 12 12:47:01 2003


Vladimir,

You could argue along the lines that interAS policy routing and
intraAS shortest path routing impose a routing structure on any wide scale
deployment of peers.  And therefore in order to capture this, one needs a
two level hierarchical model. 

See Neil and Ratul's Sigcomm03 paper or Hongsuda's work on policy
routing induced path inflation (Infocom01, I think).  You may also want to
check out a papers on cache/proxy placement, eg work by Jamin and Umich
folks etc for further motivation.  Maybe there are others on the list who
can point you to better references - I'm a bit behind on my reading :)

Anukool

On Mon, 11 Aug 2003, Vladimir Blagojevic wrote:

> Hi,
> 
> I found myself unable to effectively support the argument for choosing
> hiearchical two-level topologies for p2p simulations. In my research I
> am comparing five leading p2p solutions for internet media streaming using
> common simulator I developed.
> 
> Deployment network scenario for all these p2p streaming applications is
> internet-wide. I was influenced by all the papers that use Transit-Stub
> for the same class of research.  Therefore I just used equivalent BRITE
> model.
> 
> How do I argue that using 2-level hierarchy is appropriate for my
> research? Any references?
> 
> Most appreciated,
> Vladimir
> 
> _______________________________________________
> brite-users mailing list
> brite-users@cs.bu.edu
> http://www.cs.bu.edu/mailman/listinfo/brite-users
>