Re: [quicwg/base-drafts] Improve ACK_ECN frame encoding (e.g., use bit-vector) (#1439)

[MikkelFJ <[e-mail]>]

On the more general question of receiver duplicate tracking: yes I agree that not tracking duplicates probably opens up a lot of vulnerabilities and places an additional burden on the implementation (though a hostile peer could just repeat frames in new packets, so some robustness is required regardless).

That said, duplicate detection is a major burden of the receiver because packet numbers can in principle have arbitrary gaps so a hash table is not safe, a lookup table will not cover the full range, and a binary tree/range tree can consume a lot of memory if every second packet is a gap.

By allowing duplicates to slip through, a duplicate filter can be made much simpler, for example by hashing the last N packets and dropping any hash table collisions, or use any other simplified robust imprecise detector.

The sender generally has more book-keeping, but it knows exactly which packets it sends can therefore optimize its data structure accordingly. Only the ACKs add some uncertainty.

-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/quicwg/base-drafts/issues/1439#issuecomment-397258455