Re: [arch-d] Call for Comment: <draft-iab-protocol-transitions> (Out With the Old and In With the New: Planning for Protocol Transitions)

[Stewart Bryant <[e-mail]>]

A really good example of a transition mechanism that is not called out 
in this draft is the RIB.

In routing we have for years used the RIB as a clearing house to allow 
the routing protocols
which run ships in the night and are incompatible with each other to 
exchange connectivity
information between each other.  We can thus introduce new local routing 
protocols at will.
The (unsolved) routing equivalent of a problem of IPv6 proportions is BGP.

An example from ancient history outside the IETF was the DECnet Phase 4 
to Phase 5
transition. This was required to have a transition method indeed a as 
recall a hands-off
transition method. It was designed about the same time as IPv6. The word 
at the time
was that migration design too 80% or so of the design effort. It is 
likely that if IPv6
which presumably would have been fundamentally different, had stated 
this as a day one
goal we would be in a better place now.

Where this document appears to stop short is to recommend, maybe even 
require that all
revisions to Internet scale protocols MUST include a documented seamless 
migration
technical design, and that all new protocols that are expected to be 
deployed at that scale
MUST include the capability of being seamlessly replaced.

Best Regards

Stewart