Re: Timers and Timeouts in the SSH Transport Protocol

["Phillip Remaker" <remaker%cisco.com@localhost>]

> However, please notice that in
> numerous telecommunications protocol messages are retransmitted
> although a reliable transport level is used.
> Example: H.323/H.225.

This is really an unfair comparison, and probably a glaring exception.  The
issue with H.323/H.225 is that the protocol is frequently gatewayed to an
alternate network, and the end-to-end retransmission is required because of
errors possible after the end of the TCP tunnel/stack.  H.225/H.323 tries to
emulate telephony network protocols in a TCP universe, and since it may not
be TCP end-to-end, the retransmissions are required.

For communcations that are directly stack-to-stack, TCP should be relied
upon for retransmission.  The retransmitting behavior of H.323 is an
artifact of the legacy of the protocol.  No natively TCP based communication
should retransmit at the application layer.

It seems a bug that the identification string was not received.  The SSHD
should be able to rely on the TCP stack for reliable delivery.

I can't think of any other examples of protocols that retransmit over TCP.