Why would it be faster to hop SSH with pseudo-tty versus forwarding?

Question

I have noticed that on my office's network, multi-hop SSH connections using the "pseudo-tty" method perform better (see last note) than using the "forwarding" method. Why would this be the case?

In other words, why would:

ssh -A -X -tt server1 ssh -X -tt server2  # pseudo-tty

result in much better performance than:

ssh -o ‘ProxyCommand=ssh -A -X -W %h:%p server1’ -X server2  # forwarding

?

Other notes:

• I'm on my laptop, and server1 and server2 have the same hardware.
• On my laptop, I've got Windows 7 with Cygwin's ssh and xwin. It has a decent processor (i7 4610M or similar).
• Both server1 and server2 are RHEL 6.5.
• I haven't measured bitrate, but X11 apps are significantly faster with the pseudo-tty method (there's visible delay with forwarding).

Answer 1

I think your first command-line will estalish a connexion to server1 optimized for interactive session (with lowdelay response, but less troughtput), but the forwading method will open a session with ToS bit set to MaximizeTroughtput, with higher response time. Thus less optimized for interactive sessions.

That may come from the fact that ssh and sshd will automatically set the ToS as per rfc1349. The default is lowdelay for interactive sessions and throughput for non-interactive (usually sftp, scp, in your case port forwarding).

Extract from man ssh_config:

 IPQoS   Specifies the IPv4 type-of-service or DSCP class for connections.  Accepted values are af11, af12, af13, af21, af22, af23, af31, af32, af33, af41, af42, af43, cs0, cs1, cs2, cs3, cs4, cs5, cs6, cs7, ef, lowdelay,
         throughput, reliability, or a numeric value.  This option may take one or two arguments, separated by whitespace.  If one argument is specified, it is used as the packet class unconditionally.  If two values are speci‐
         fied, the first is automatically selected for interactive sessions and the second for non-interactive sessions.  The default is lowdelay for interactive sessions and throughput for non-interactive sessions.

ps: you may try:

ssh -o ‘ProxyCommand="ssh -A -X -W %h:%p -o IPQoS=lowdelay server1"’ -X server2