> > Note that NTS uses certificates to authenticate the KE. Generally, a
> > certificate chain cannot fit into a single IP packet.
>
> Yes, there was a discussion about that on the ntpwg list recently and
> currently NTS relies on IP fragmentation even with the standard
> ethernet MTU of 1500. I agree it could be a problem. It seems IPv6
> doesn't even require hosts to defragment packets larger than 1500
> octets.
>
> Adding fragmentation to the NTP protocol seems like a horrible idea.
To be honest, we had never assumed for IP fragmentation (on one request
and one response) to be a huge problem.
If people see grave problems with it, we would welcome an elaboration as
to why this is.
> Would it be safe to assume an NTP+NTS packet can always contain a
> single certificate without requiring fragmentation with MTU of 1500?
> Perhaps the server_cook message could be split into multiple messages.
> With the client_cook message I suspect it would be much more
> difficult.
Just for the record: as per the current draft version, the messages in
question (those which likely exceed MTU because they contain certificate
chains are server_assoc and client_cook).
To clarify: does the reason you suspect that splitting client_cook is more
difficult lie only in the stateless nature of the server and the fact that
client_cook is a request, sent from client to server? Or am I missing
something?
Kristof
_______________________________________________
TICTOC mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/tictoc
