Andrew Mcgregor <[email protected]> wrote:
> 
> Andrew McGregor: If you find yourself in that situation, there are
> approaches. In exists deploments where you have a different ECN inside the
> DC then outside torward the internet.
> 
> Should read something like "If you're trying to enable ECN in a large
> content provider, you can find the routers that cause you problems and
> disable ECN toward the subnets behind them.

   (I'm not arguing that you didn't say that...)

   "Finding" such routers isn't trivial. :^(

   Identifying subnets behind them is even less trivial. :^(

> It can be a more serious issue to deal with mark-equals-drop ECN on
> the outside and early-onset ECN on the inside, since traffic often
> crosses the same switches in both cases."

   I wish you'd find a clearer way of saying this -- in some follow-up
thread on this list. I am left scratching my head as to what you're
suggesting we _do_ about this (even if these cases were easy to
distinguish).

   I would like to simply enable ECN on a mostly-vanilla webserver;
and use packet-filtering _if_ I learn of a complete disaster somewhere:
otherwise letting the chips fall where they may.

   IMHO, a router which simply clears ECN isn't a problem I need to
worry about. Even a router which drops CE-marked packets isn't worth
worrying about for a vanilla web-page.

   OTOH, a router which drops all ECN-enabled packets is major damage
which needs public humiliation. (So I'm not in a big hurry to hide
its broken behavior.)

   This issue clearly deserves discussion (though I'm not in any
hurry to convert others to my views). Perhaps there is a better list
for such discussion, but it's not obvious where else to hold it...

--
John Leslie <[email protected]>

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

Reply via email to