On Fri, Jul 26, 2024 at 8:17 PM Alia Atlas <[email protected]> wrote:
> I read draft-fenner-intarea-extended-icmp-hostid-02 and think it is a very > useful idea. > Thanks! I'm thinking about it also from the ability to provide additional > abstraction information that could be useful to annotate a node with. > > For example, in a Cloud context, knowing the region associated with a > router is useful and the Host Identification Object might allow providing > appropriate abstractions. > > Would it be of interest to consider using one of the reserved flags in the > Ctype of the Node Identification Object to allow additional information? > > I'm thinking of something where there might be the ability to allocate > with a document or an organization field and then different meanings > allocated there. > This is going to be a pretty popular idea, I think - it came up in the WG session too. There are 6 bits left; one idea is to keep this document specifying just IP address and name, and add a more complete specification of the details of how to encode a value with a later bitfield allocation, and define the IANA rules for such an allocation. If we do take on the task of specifying how more information types get included, I am concerned about falling into a bikeshed argument about exactly what information should be able to be included, and what format those should take, etc., which would delay the eventual publication of the current spec on which I think we have a fair amount of agreement. I fear that a specification for "and here's some more data of an arbitrary format" will have to include a type, and a length, and rules about how to skip TLVs that you don't understand - and at that point, why do we not just define a new RFC4884 extension type and use RFC4884's type and length and rules about how to skip TLVs that you don't understand? If there's some concrete, well-defined information that we think we can come to consensus on - with respect to both usefulness and format - then I'm all for adding it to this document. And I wouldn't be against a length-limited text field with a relatively abstract name - "annotation"? - if there's WG support for it. Bill
_______________________________________________ Int-area mailing list -- [email protected] To unsubscribe send an email to [email protected]
