It often seems like all these tunnel types serve individually useful
purposes, but that we've also simply failed to reuse much if anything,
making implementation more difficult. Its seems, to me, that could
live with a few generic tunnel types, rather than a thousand
specialized ones, but the cat already seems to be out if the bag (?).
we used to repeat a mantra along the lines of reliable ip transport
reinvention only shows that tcp (this year's version, of course <g>) is
sufficient and a good compromise among the many constraints.
that philosophy never took hold in the encaps space. a tecnoloogy for
every need. how much was real need and how much career/market building
is left to the judgement of the observer.
randy