Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!mnetor!uunet!husc6!cmcl2!nrl-cmf!ames!ucbcad!ucbvax!AFWAL-AAA.ARPA!west%etd1.DECnet From: west%etd1.DECnet@AFWAL-AAA.ARPA ("ETD1::WEST") Newsgroups: comp.os.vms Subject: I forget Message-ID: <8711220544.AA05679@ucbvax.Berkeley.EDU> Date: Mon, 16-Nov-87 18:29:00 EST Article-I.D.: ucbvax.8711220544.AA05679 Posted: Mon Nov 16 18:29:00 1987 Date-Received: Tue, 24-Nov-87 00:49:25 EST Sender: daemon@ucbvax.BERKELEY.EDU Reply-To: "ETD1::WEST" Organization: The ARPA Internet Lines: 55 From: AVLAB::MAILER 16-NOV-1987 14:31 To: ETD1::WEST Subj: DDN Mail Message Return-Path: <@wiscvm.wisc.edu:POSTMAST@DTUZDV5A.BITNET> Received: from wiscvm.wisc.edu by afwal-aaa.arpa ; 16 Nov 87 13:29:48 EDT Received: from DTUZDV5A.BITNET by wiscvm.wisc.edu ; Mon, 16 Nov 87 13:30:08 CDT Date: Mon, 16-NOV-1987 18:41:43.21 GMT+1 From: Subject: Returned Network Mail To: Your mail is being returned to you. Reason for return is: %MAIL-E-NOSUCHUSR, no such user Z112742 at node ZMVAX Returned mail follows: ------------------------------ Received: From DEARN(MAILER) by DTUZDV5A with RSCS id 9153 for Z112742@DTUZDV5A; Mon, 16 Nov 87 18:41 N Received: by DEARN (Mailer X1.24) id 9150; Mon, 16 Nov 87 18:40:12 CET Date: 13 Nov 87 08:36:00 EDT Reply-To: "ETD1::WEST" Sender: Info Vax From: "ETD1::WEST" Subject: LAT vs. any other kind of terminal connection Comments: To: info-vax Comments: cc: labovitz%etd2.decnet@afwal-aaa.arpa To: Max MAIER At the Air Force Avionics Lab, we have an Ethernet backbone with (at least) 12 VAX and lots of graphics workstations attached. This backbone runs DECnet, LAT, CALMAnet, TCP/IP and more. This backbone serves AT LEAST 40 and maybe more highly utilized terminal servers (DECserver 100 and 200's.) This is 320 terminals and PC's connected by LAT, there are also several (10-15) devices connected by reverse LAT on DS200's...FOM's etc. My point...the net is busy but not strained. There is NO terminal I/O degredation...and this technique beats using DZ-11's hands down for off loading processing. A performance audit showed that 15% of the CPU was being spent handling interrupts (from DZ-11's). After getting rid of the DZ's and going to LAT, the time spent in interrupt was .LT. 2%. Over 10% of the VAX cpu time was returned to the users for the price of an upgrade which is justifiable in terms of convenience (for distributed connect capability) alone! Mike West Air Force Avionics Lab AFWAL/AADE WPAFB, OH 45433 west%etd1.decnet@afwal-aaa.arpa ------ ------