On 03/19/2016 11:03 PM, Marrold wrote:
That's essentially what I'm hoping for, although I appreciate from an RFC purist perspective it's probably a bit of a nasty hack.
It's not a question of pedantic RFC purism so much as:
(a) Whether it creates at least as many complications as it solves; solutions like this one create so many complications, corner cases and statekeeping requirements that they represent - to borrow a term from philosophy of science - a "degenerative heuristic". This means crawling down a rabbit hole to solve the wrong problem the wrong way and just getting tangled.
But perhaps a more applicable term would be:
https://en.wikipedia.org/wiki/Anti-pattern
(b) Whether it's truly the most effective solution to the problem. It seems to me you're solving the wrong problem; solutions for UDP fragmentation (use TCP, remove other, less essential headers) and topology hiding (B2BUA) are well-understood and canonical.
(c) Technically possible.
-- Alex