Re: [quicwg/base-drafts] Preventing KEY_PHASE bit from being used as a tool to correlate CIDs (#1322)

[Martin Thomson <[e-mail]>]

Key updates require reciprocation.  Using a new connection ID would be the same.  So no change there.

One advantage of coupling NEW_CONNECTION_ID to key updates is that it means that there is a single mechanism that supports migration (and multipath) and key update.  You can't support one without also enabling the other.

One thing I've been concerned about with key update is that no one will implement it.  NSS only very recently grew that capability in TLS 1.3; it's a small feature, but one that took ages to implement because there wasn't much incentive to build it.  We still don't have the DTLS variant because it messes with our timers in ways that are hard to reason about.

I agree that simultaneous migration would be a mistake.  Better to drive rollovers based on packet counts as @kazuho suggests.  Though I think that we can probably ignore the 2<sup>62</sup> limit and throw connections away if they ever get to that point.

-- 
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/1322#issuecomment-385601496