IESG proposed statement on the IETF mission
Eric Rosen
erosen at cisco.com
Tue Oct 21 12:32:13 CEST 2003
> Why an IETF standard, and not a Traffic Light Controllers Forum standard?
Here's the typical progression.
The "traffic light controllers forum" concludes that the Internet doesn't
really provide adequate service/qos/oam/security or whatever and that
controlling traffic lights requires an ATM/FrameRelay/DialUp/X.25/Ethernet
or whatever. (Perhaps the rules for voting in the traffic light controllers
forum prevent the technical issues from being considered.)
A few rebel traffic light providers then get together with one or two
networking vendors and figure out how to do the traffic light control over
the Internet.
In a short time, this proprietary mechanism becomes widely deployed, while
the "standard" produced by the forum has no deployment. The forum fails to
see this, and continues to work on perfecting the standard by adding
additional OAM to it.
The rebels (who by now are controlling most of the traffic lights) start to
worry about their dependence on one or two vendors, and decide they want a
published standard with some sort of imprimatur. So they ask for an IETF
WG.
Meantime, the IAB (whose job is to fret and deplore) issues a paper
deploring the inadequate congestion control, security, and scalability of
the proprietary mechanisms, fretting about the effect on the Internet.
Prominent IESG members announce that anyone who pays attention to traffic
lights is clueless, that no one will ever control them over the Internet
anyway, and that the whole thing is just a vendor scam. Operators don't
care whether traffic lights work or not, controlling them just adds
complexity. Besides, traffic light control protocols weren't needed in
1990, so how can they be needed today?
A WG is formed, but the charter requires that a requirements and framework
document be produced before any protocol specification work is done.
Four years later the IETF standards for the application get approved. They
are quite similar to the proprietary mechanisms, which already dominate the
industry. Prominent IESG members still insist that no one will ever deploy
them. Other IETF members ask why the standards have been rushed through.
The traffic light controller forum then endorses the IETF standard, renames
themsevles to be the "IP-based real time device control forum", and asserts
itself to be the proper standards body for all kinds of IP-based protocols.
Seriously:
- A considerable amount of time could be saved if the IETF just does the
work in the first place.
- The very last thing we want is a proliferation of clueless forums claiming
to be producing standards for IP-based protocols; it's hard to see how
that is "good for the Internet".
- Traffic light control was picked as an example in order to make IETF
involvement look absurd; if a more realistic example (providing some sort
of network service or network infrastructure) had been picked, the proper
conclusions would be even more obvious.
More information about the Problem-statement
mailing list