[c-nsp] Routing design question

[Marian Ďurkovič]

On Tue, Oct 02, 2007 at 12:01:30PM +0200, Gabor Ivanszky wrote:
> Marian Ďurkovič wrote:
> > On Thu, Sep 27, 2007 at 09:56:06AM +0200, Gert Doering wrote:
> >> Will auto-neg signal one-way fiber failures (after the link has
> >> already been brought up and autoneg'ed successfully)?  Never
> >> tried that.
> >
> > Yes, it works much better (and faster) than UDLD. The endpoint
> > which has lost receive fiber will immediately signal "RemoteFault"
> > to the other end and line protocol will go down on both sides
> > within miliseconds.
> >
> 
> true, but if they have some active devices between the two sites(e.g.
> an SDH/SONET transmission facility owned by the telco providing the GE
> service), those active devices could fool the customer's GE interfaces
> making them believe the connection is up and running, however the
> circuit may or may not be operational.

Well, even this case could be handled by RemoteFault indicator - if
SDH/SONET devices were designed accordingly.

For example, if there's some fault in the SDH backbone, every SDH box
signals AIS to E1 tributary interfaces. It could use the RemoteFault
indicator on ethernet tributaries in the same way - causing the customer's
interfaces to go down immediately.

The only annoyance here is that IOS is not currently able to deliver
RemoteFault info to the user - the interface just shows "down".
But we have a switch from other vendor which is clearly stating
"Remote fault" in this case.

Of course, BFD and/or dynamic routing protocol are still necessary
in more complex scenarios.


	With kind regards,

		M.