Rob, thanks
  > I'm not a hard-core ssh expert either, but keep in mind that we're
  > already using the ssh "subsystem" mechanism, ie, we are using an
  > assigned name for the specific service we're trying to reach.
sorry, I forgot/missed that use of ssh subsystem "rpki-rtr" is already
in the draft; that clearly does the selection of application service.

  > Furthermore, while outside the network spec, implementations like
  > OpenSSH already have the concept of restricting a particular set set
  > of authentication credentials to running a command determined entirely
  > by the server.  That is: ie, it is not difficult to lock a particular
  > set of credentials into running only the specific command or subsystem
  > one wants to allow.  Small matter of configuration, but one would have
  > to configure credentials in any case, this is just a bit more config.
  > 
  > I think the port number thing is more about people who really just
  > want to be able to filter based on port numbers, full stop.  I don't,
  > particularly, because forcing encrypted traffic (content the filter
  > can't inspect, by definition) to use port A instead of port B has
  > never struck me as an effective policy enforcement mechanism.  YMMV.
With application selection taken care of elsewhere using seperate addresses
for helping packet filters etc. seems perfectly appropriate;
using separate well defined port would be wasteful.

Ruediger Volk
_______________________________________________
sidr mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/sidr

Reply via email to