[Xorp-users] RTT "Round Trip Time"

[TTA Group]

Thank you guys for your reply
I will take your advice about not playing with RIP to make it
RTT(Round Trip Time). But it seems XORP does not support any of the
"Cost" link protocols like OSPF or EIGRp isn't it ?

So in this case, either I have to install their routing platform (like
Zebra) or even change the OS and use Windows2000 for OSPF. so which
choice you recommend guys?

I notice that to activate OSPF in windows2000 I have to configure
Routing and Remote Access service in it. and that is only available in
windows 2000 server. isn't it ? do you think I can easily configure
OSPF in windows to use the delay metric (close to RTT) ?

Thanks for your help


On 5/11/05, Gernot W. Schmied <gernot.schmied@chello.at> wrote:
> Pavlin Radoslavov wrote:
> >>I'm a graduate student and my graduate project about "Round Trip Time"
> >>as router matrix. So I think to use XORP as my research environment.
> >>But I notice that the default xorp RIP matrix is "Hob Count". So does
> >>any one has any idea how I can make it RTT "Round trip time"? any help
> >>or suggestions in this subject or previous experience?
> >
> >
> > Tariq,
> >
> > The RIP protocol itself is designed to use hop-count, so it is not
> > clear whether it can easily be modified to use RTT instead of
> > hop-count. It may be simpler to design your own routing protocol
> > that uses RTT instead of modifing RIP, but you should be the one to
> > research that option :)
> >
> > Once you have designed your routing protocol, then you can decide
> > whether you can reuse some of the RIP code.
> > If you want to implement your protocol from scratch, then you can
> > use the static_routes implementation as a starting point.
> >
> > Good luck!
> > Atanu & Pavlin
> > _______________________________________________
> > Xorp-users mailing list
> > Xorp-users@xorp.org
> > http://mailman.ICSI.Berkeley.EDU/mailman/listinfo/xorp-users
> >
> 
> Designing a routing protocol around a volatile, largely varying and
> intrinsically unpredictable metric such as RTT is a difficult task and
> generally not a good idea in a best-effort IP network. I'd rather
> suggest you take an existing routing protocol such as OSPF or EIGRp and
> consider implementing the generic "cost" hook based on consolidated RTT
> data instead of bandwidth, hop-count or something else. Anyway, you have
> to consider the return-path as well so that would defeat the purpose.
> For a start you could also look at the EIGRP hybrid metric approach.
> 
> Regards,
> Gernot
> _______________________________________________
> Xorp-users mailing list
> Xorp-users@xorp.org
> http://mailman.ICSI.Berkeley.EDU/mailman/listinfo/xorp-users
>