draft-rfced-info-banan-00.txt

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

I do not consider this document fit for publication as an RFC
in its current form.

The protocol itself is poorly designed in a number of areas.
It might be more efficient than SMTP in terms of number of
packets and delay, but on a quick review it appears to be
gratuitously complex, lack important functionality, and have
insufficient error reporting and diagnostic capability.
Its design ignores much that has been learned about email over
the past 20 or so years.

The use of ASN.1 and ESRO are highly questionable.  ONC RPC/XDR 
and T/TCP would appear to be far better choices.  

The document also contains a number of inaccurate characterizations 
of Internet mail and SMTP.

If this is indeed widely deployed, then the community might benefit
from publication of this document as an Informational RFC, but it
should first be edited to remove these mischaracterizations and/or
have a strong disclaimer that recommends against actually using it.

There is a real need that the protocol attempts to address, as SMTP 
is not well suited to an environment where there is a high cost 
associated with transmission of each IP datagram (it uses more 
datagrams than absolutely necessary), or where there is a long 
round-trip delay combined with low reliability (the several round
trips required to complete an SMTP session can cause considerable 
delay in message delivery).

So we should probably consider doing a better design in IETF, but
it should not be based on this protocol.

Keith