Re: [quicwg/base-drafts] Connection ID Length changes (#2473)

[Kazuho Oku <[e-mail]>]

@martinduke IIUC, generally speaking, trial decryption becomes inevitable for a client when it coalesces multiple connections that uses the same server 2-tuple onto a single socket.

In the approach you describe, the client is not coalescing multiple connections in such a way on a single socket when it initiates the connection, because it is initially using zero-length CID. But it sounds like that the client will be using the socket in such a way on a socket it decides to migrate.

What I do not see is the rationale behind such a design.

If the client's preference is to minimize the network bandwidth (by using zero-length CIDs), shouldn't it use zero-length CIDs for subsequent paths that it migrates the connection to? That's possible, as long as it does not coalesce multiple connections that share the same server 2-tuple onto a single socket. The simplest way of doing that is to use a dedicated socket for each connection.

Or if the client's preference is to minimize the number of sockets it maintains, shouldn't it be always using non-zero-length CIDs from when it initiates a new connection?

-- 
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/2473#issuecomment-464425588