Les,

 
> Nice acronym. 😊


Thank you. I blame the wordle craze. :)


> As regards the original use case (Node Liveness), my opinion hasn’t changed. 
> Repackaging this in a more generic wrapper (which makes sense to me) doesn’t 
> alter my opinion – which is this isn’t a good way to go.


This is well known and does not need restatement.  :)


> As regards the new use case (Capabilities), I think what you state as 
> historical behavior isn’t accurate. If you look at the existing sub-TLVs for 
> Router Capability, the information falls into two categories:
> Stuff directly used by the protocol
> TE related stuff (e.g., PCE)
>  
> I would oppose any attempt to move stuff directly used by the protocol out of 
> Router Capability and force the IGP to get this info from some application.


And that’s not what I was suggesting.

I am NOT suggesting we move anything that is currently defined, regardless of 
whether it is appropriate or not.  I would simply like to avoid forthcoming 
atrocities.


> Which leaves stuff NOT used by the IGP. The only existing example which 
> “might” fall into this category is: S-BFD Discriminators – which we stuck 
> into the IGP for lack of a better place.
> I therefore think your language regarding what is/is not a good candidate for 
> DROID capability advertisement needs refinement. As written, it suggests that 
> there are lots of existing cases that are being advertised by the IGP 
> inappropriately – which I do not think is the case.


Well, we can continue to disagree.

I was, in fact, thinking of the slew of capabilities that are about to be 
proposed for the MPLS MNA work.  I know of about ten new capabilities just from 
that direction.


>  I also think that asking the IGP to advertise the location(s) of DROID is an 
> abuse of the IGP – the very thing that you have argued should not be done. 
> The IGP – and specifically Router Capabilities – is not intended to serve as 
> a form of DNS – advertising the location of application services in the 
> network.


Yup. One small sin to prevent a host of subsequent sins.


> It is ironic to me – given that you have framed DROID as motivated by  
> “anti-IGP-as-a-dumptruck”  – that you would then abuse the IGP by asking it 
> to advertise the location of DROID. It opens the door to others who want the 
> IGPs to advertise application stuff – but when that is rejected by LSR WG (as 
> it should be) they say “OK – but can we advertise the identity of a place to 
> get the info in Router Capabilities?”.
> This is unjustifiable IMO – please remove it.


I will be happy to do as soon as there is a working service discovery protocol. 
 But I’m not holding my breath.  Until then this is expedient.


> Finally, there is the question of where such a draft belongs. LSR seems like 
> the wrong place.
> Where are you headed with this?


The LSR chairs have said that this is an appropriate home for now. There is no 
point in discussing this.  Nothing has changed.

Tony

_______________________________________________
Lsr mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/lsr

Reply via email to