The MPLS WG Archive

[tom worster <fsb@thefsb.org>]

it seems that verification at the decapsulator and source address filtering
at border routers are both valid solutions and there could be reasons for
preferring the one or the other in different operational scenarios. it isn't
clear to me that either way is an obvious winner.

as you pointed out, eric, requiring verification at the decapsulator could
involve a design complexity cost even if the function is not actually used.
so demanding that all decapsulators provide that function would be an
unnecessary cost if there are significant numbers of operators who would not
use it.

if the options for this draft are:

a) specify the decapsulator verification _and_ require it

b) specify verification at the decapsulator as an implementation option,
i.e. specify it and explain its applicability but do not _require_ it.

c) neither specify it nor require it

and if we don't observe a rough consensus on whether filtering at the border
or at the decapsulator is preferable in this forum then i would think that
b) is the way to go for the time being and we can hope that a rough
consensus may emerge as we progress towards a draft standard. in general i
dislike putting options into protocol specs but in this case it may be the
right thing to do.