At the OSPF WG meeting in Chicago, Stephane Litkowski expressed the desire to
have a more dynamic mechanism for binding attributes to applications. I’ve
thought about this for some time and see the following potential use cases.
1. Usage for an application prior to early IANA allocation of a bit
position for that application.
2. Usage for user applications that will never be standardized but still
require link (and potentially prefix) attributes.
Are there any others?
An application tagging mechanism was suggested for this purpose. However, I
still believe the bit-mask encoding can satisfy these use cases.
The first requirement could be satisfied by allocating some number of
experimental bit positions in the initial bit mask specification. These could
be used for experimental applications prior to IANA early allocation.
Similarly, the second requirement could be satisfied by allocating some number
of user-application bit positions in the initial bit-mask allocation. I really
don’t see significant advantage of a tag over a bit position as, no matter what
mechanism is used, all the OSPF routers in the routing domain will need to
agree on what bit/tag corresponds to what application. For the first use case,
this would be decided among the implementors supporting the experimental
application while in the second use case, the application binding would likely
need to be configured.
The only argument I could see for a application tag would be that you need too
many bits of either type (experimental or user applications). In this case, we
could still use a bit mask in the OSPF Prefix/Link Attribute LSA and advertise
the application identifier (i.e., tag) to bit position mappings in the OSPF
Router Information LSAs. However, I don’t think this level of indirection is
necessary.
Thanks,
Acee
_______________________________________________
OSPF mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/ospf
