Hi Tony, Gunter,
On 02/03/2022 16:22, Tony Li wrote:
Hi Peter,
I believe that your assumption is that the FAD for a single algorithm can never
grow bigger as a 256 octets.
yes, that is indeed a limitation.
Perhaps we should address it somehow? Coders have to deal with overflow
conditions on day one, whether people are hitting them yet or not.
If FlexAlgo is adopted, then we should expect that it will get stressed and
that further conditions will be added to a FAD. The problem will only become
worse the more success that FlexAlgo has. It sounds to me like you want
FlexAlgo to fail, which seems strange.
I believe there is a subtle difference between what each of you is saying:
a) Tony says that there may be many constraints used for a particular
flex-algo, so that we may not be able to fit them in a single FAD
Sub-TLV. I agree, we better address that. I will update the flex-algo
draft accordingly.
b) Gunter says that a particular Sub-TLV of the FAD Sub-TLV, on its own,
may not fit into a single FAD Sub-TLV. Although theoretically possible
(with SRLGs only for now, I believe), I don't feel like it's very realistic.
I agree to address (a), not very enthusiastic about addressing (b).
Please let me know what do you think.
thanks,
Peter
What if the FAD use SRLGs? Each of these SRLGs are big (4 octets) in size and
will consume significant subTLV space resources and many of those may overrun
the size of the FAD. So if we have a theoretical network with a theoretical
use-case that wants to exclude, lets say 64, SRLGs, then how would that fit
into a single FAD for a single algorithm?
you would not be able to do that. Realistically, it looks like an academical
problem to me.
Actually, what’s academic is your objection. Implementations are going to
handle overflows and they will do it by generating multiple sub-TLVs. The
problem is not going to go away just because you deny that it exists.
Anyway, my point was that draft-pkaneria-lsr-multi-tlv may benefit to call-out
subTLVs that by normative reference can only exist one time, in one place, like
for example the flex-algo FAD.
I don't believe draft-pkaneria-lsr-multi-tlv should go into business of
specifying the restriction of any particular Sub-TLVs. These restrictions are
specific to Sub-TLV itself and should be covered by the specification that
defines the Sub-TLV.
Except that many things didn’t define how they can and should be extended.
Admittedly, that’s a bug in those specifications that we’re attempting to
address.
Tony
_______________________________________________
Lsr mailing list
Lsr@ietf.org
https://www.ietf.org/mailman/listinfo/lsr
