To:
ggm@apnic.net (George Michaelson)
Cc:
bmanning@isi.edu (Bill Manning), peter@interq.or.jp (Peter Chow), yu.zhu@i-dns.net (Zhu Yu), ietf-provreg@cafax.se
From:
Bill Manning <bmanning@isi.edu>
Date:
Wed, 21 Mar 2001 13:25:02 -0800 (PST)
In-Reply-To:
<15892.985206638@apnic.net> from "George Michaelson" at Mar 22, 2001 06:30:38 AM
Sender:
owner-ietf-provreg@cafax.se
Subject:
Re: Design teams
% % % If you use UDP, you have to roll your own VC to try and leverage anything % from connected state. Re-inventing the wheel is usually bad. Yup. % If you are looking for transactional completeness, I suspect UDP/TCP is % the wrong question. However you do get a little more help to know if data % made it without neccessarily needing an applications-layer ACK. % % maybe? figure that most transactions will use some transport protocol and the transport protocols will eventually run on either UDP or TCP. Stealing an existing transport protocol is easier than writing one from scratch. % % -George % -- --bill