Re: I-D ACTION:draft-shen-bfd-intf-p2p-nbr-00.txt

[Naiming Shen <[e-mail]>]

Nitin Bahadur said the following on 03/30/2007 04:41 PM:
> Not sure of the exact problem you are trying to solve. Instead of using
> BFD as a link failure detection mechanism for Ethernet, can't you use
> Ethernet OAM? If this draft is a stop-gap mechanism before Ethernet OAM
> is implemented/put-in-use, maybe it's ok.

If Ethernet OAM works for you, yes use it. This draft is not just designed
for Ethernet though.

> 
> To implement the draft correctly, a lot of special case handling would
> need to be added in multiple places. It would be better to use a
> link-layer mechanism for detecting link failures.

You mean bring down and bring up and intf is more work than putting hooks
everywhere in every protocol and every service? not to speak for every
implementations, for certain ones, one and half day of coding and testing
should cover this draft.

> 
> Also, with the draft you are tying in the concept of a link failure to a
> bfd session failure...which might not necessarily be true. BFD sessions
> might fail due to firewall filters, IP packet handling errors. You would
> need more tools to tell the customer/operator that the link is *actually
> not down* and it's BFD that has marked the link as down.

to be honest, generate an meaningful error message saying 'intf-p2p bfd
brought down intf' will probably do for most of the customers.

thanks.
- Naiming

> 
> Nitin