Christopher Morrow wrote on 04/12/2018 20:17:
I think the point that Nick and Gert are outlining is that if the case is that the hardware available will have no fast-path processing for packets with obtuse patterns or sets of extension headers those packets will get sent to the control-plane (slow-path).
in this case, yes, but there is a more general issue, namely that protocols are often written without the benefit of hindsight or experience. Implementation of specifications will usually throw up problems and it almost never happens that a protocol specification is defined so jesuitically that ambiguity doesn't happen.
Looking at this from another angle, engineering is about optimisation of conflicting input parameters. The IETF needs to understand how and why this happens, because it can provide useful feedback on the rfcs that are churned out, ranging from ambiguity to gap analysis to impracticality.
Nick
