Re: [OSPF] Asymmetric OSPF Hold Timer draft

    Hi Russ,
    there is no harm in sending Hellos more frequently than advertised 
Hello interval.
    In other words, if you look at the problem from neighbor discovery 
angle alone (i.e. state that neighbor liveliness detection is offloaded 
from Hellos) then it is best to set Hello and Dead intervals to [near] 
infinity and send Hellos as dictated by other requirements. OSPF could 
send Hellos as fast or as infrequent as needed for neighbor discovery 
and efficiency; symmetric, asymmetric; behavior the same on connected 
routers or different etc.
    If implementation wants to do asymmetric neighbor discovery then 
this is doable within the scope of 2328.

Anton


On 03/24/2013 01:31 PM, Russ White wrote:
>> BFD was, among others, design for that. Removing it is bot an option in my opinion. Operators tell us vendors to make cost-effective solutions, then request functionality extensions that make control plane more complex because they try yo support today's best functionality without the tools invented for this. I think expanding OSPF to support what we can do already is not the right thing to do.
>
> My impression of this draft is that it's not about fast down detection
> (as BFD is), but asymmetric discovery speed to improve efficiency. Can
> you point me to the place in the BFD draft that covers this capability?
>
> The question isn't whether or not BFD can do this particular thing --the
> questions are:
>
> 1. Is this a reasonable thing to want to do?
> 2. Should we extend OSPF to do it?
>
> This isn't the BFD working group (?).
>
> :-)
>
> Russ
>
>
> _______________________________________________
> OSPF mailing list
> OSPF@ietf.org
> https://www.ietf.org/mailman/listinfo/ospf
>