Received: from localhost by CS.UTK.EDU with SMTP (cf v2.9s-UTK) id EAA05021; Tue, 5 Mar 1996 04:01:29 -0500 Received: by CS.UTK.EDU (bulk_mailer v1.4); Tue, 5 Mar 1996 03:59:24 -0500 Received: from rome.software.com by CS.UTK.EDU with ESMTP (cf v2.9s-UTK) id DAA04747; Tue, 5 Mar 1996 03:59:06 -0500 Received: from rome.software.com ([198.17.234.100]) by rome.software.com (post.office MTA v1.9.1 ID# 0-10001) with ESMTP id AAA18689; Tue, 5 Mar 1996 00:58:41 -0700 X-Mailer: exmh version 1.5.3 12/28/94 To: djb@koobera.math.uic.edu (D. J. Bernstein) cc: drums@cs.utk.edu Subject: Re: proposed agenda for 8 March WG meeting In-reply-to: Your message of "05 Mar 1996 07:17:44 GMT." <19960305071744.3858.qmail@koobera.math.uic.edu> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Tue, 05 Mar 1996 00:58:37 -0800 From: michael.derrico@software.com (Michael D'Errico) Message-ID: <19960305085840.AAA18689@rome.software.com> > I meant exactly what I said: except in extreme circumstances, good MTAs > can achieve much lower latency with separate SMTP connections than with > multiple RCPTs. sendmail can't, but my MTA, qmail, can and does. Your mostly-unused 10 Mbps LAN is probably not a good testbed for your theory. It's much easier to saturate a T1 line, especially when it's shared with a WWW server handling several hundred thousand hits a day. PIPELINING is the solution for latency problems in SMTP. It requires fewer round trips than plain-old-SMTP, and is a big improvement even in the case of a single recipient. I suggest you focus your energy on implementing it. Mike