Re: Last Call: IP over MIME to Proposed Standard

[Keith Moore <moore@cs.utk.edu>]

IMHO, this is totally unacceptable for Proposed Standard status.

1. it recommends using MIME, and presumably SMTP, as a means of
tunneling IP packets.  Even considering that a content-type for
transmitting IP over MIME might be useful for "monitoring, analysis,
debugging, or illustrative purposes", that doesn't mean that we should
standardize the use of this content-type for tunneling IP packets in
operational use.  Neither the SMTP protocol nor SMTP implementations are
designed for this purpose, and the delays and jitter involved in sending
IP over SMTP make it marginally suitable at best - certainly not
something that can be expected to scale, or to work satisfactorily over
most of the Internet.  In other words, this proposal would fail to meet
criteria that we routinely apply to standards-track documents.

Furthermore, this document makes no attempt to narrow its applicability;
rather, it specifically claims that use of IP-over-MIME as a tunneling
mechanism is within scope.

2. the Introduction recommends use of NAT, which is nonstandard and
known to often be a harmful practice.  

3. Several references cited as Informative are actually Normative -
in that you can't implement the IP over MIME protocol without them.
Among these references are the reference to IETF documents which are
not on the standards track.

If we approve this as PS we are essentially recommending a stupid
practice.  At best, it's embarassing to IETF to do so; at worst, we are
endorsing an operational practice that has little practical use other
than to circumvent existing security measures.  Even if those security
measures are poorly chosen, it is not appropriate for IETF to recommend
their circumvention.

Even assuming there is a legitimate use for this, there is no need to
standardize this content-type, since content-type definitions do not
need to be on the standards track.

Keith