ietf
[Top] [All Lists]

Re: ietf.org end-to-end principle

2016-03-17 08:05:13
On Thu, Mar 17, 2016 at 6:38 AM, Stephen Farrell
<stephen(_dot_)farrell(_at_)cs(_dot_)tcd(_dot_)ie> wrote:


On 17/03/16 10:31, DIEGO LOPEZ GARCIA wrote:
 whether the good-ole e2e principle is no longer applicable

I always think of it as the end to end argument, not principle,
and from that perspective, I think it remains entirely applicable.

S.


Exactly. And the end to end argument is entirely about the decisions
you make about placing complexity.

What has not changed in 30 years is that the Inter-Network remains an
area where complexity is to be avoided. If you are working at the
packet layer, the end-to-end principle is unchanged.

Above the packet layer, the end-to-end principle was never absolute
and it can't be. A given bilateral communication can only be initiated
by one party which means the other has to be the responder. It is not
possible for an initiator to communicate directly with an initiator.
The only way such a communication can be established is if they both
talk to an intermediary that can act as a responder for both.

If you look at the design of SMTP, you will see this architecture has
been in place for 30-odd years. And it is in Jabber and pretty much
every other 'peer to peer' scheme.


In every protocol, we try to reduce complexity wherever we can. But
there is always a minimum that can't be reduced except at cost to
functionality or performance or creating complexity elsewhere.

If you read the end-to-end paper you will find arguments against
putting that complexity at the ends as well as the network core. In
many but not all cases, a middlebox is exactly where it should go.

<Prev in Thread] Current Thread [Next in Thread>