[ On Sun, November 22, 1998 at 15:25:53 (-0500), der Mouse wrote: ]
> Subject: Re: SOLVED! The cause of puzzling TCP (eg. WHOIS) connection failures with some InterNIC.net hosts 
>
> Saying that PMTU-D is broken-as-designed because it fails when hosts
> drop need-frag ICMPs is like saying TCP is broken-as-designed because
> it fails when hosts drop data-free segments (ie, pure ACK).  The only
> sense I can see in which the analogy can be considered unreasonable is
> that the former is somewhat commoner in today's Internet.

I think your analogy is completely wrong.

For starters, PMTU-D is an *optional* protocol.  That failure of an
optional protocol can cause TCP connections to fail in mysterious ways
is *bad* -- it means, at least to me, that PMTU-D is not robust and
since it is intertwined with TCP it means that PMTU-D+TCP is less robust
than TCP alone.

If, and only if, PMTU-D had required this as-yet mysterious "black-hole
discovery" feature right from the start then I think it would be
sufficiently robust.

> In short, blame the broken box, not the protocol.

I blame both in the current circumstances.

(although unfortunately in some cases it's not just one box that's
broken)

-- 
							Greg A. Woods

+1 416 218-0098      VE3TCP      <gwoods@acm.org>      <robohack!woods>
Planix, Inc. <woods@planix.com>; Secrets of the Weird <woods@weird.com>