smb@ulysses.UUCP (07/21/83)
I posted a similar query some time ago, without much success, but
I'll try again?
Are there other users of 3Com's UNET out there, who would like to share
information on bugs, fixes, etc.? I'm especially interested in talking
with folks who have networks running both UNET and 4.1c/4.2; I've run
across some *fascinating* incompatibilities....
--Steve Bellovin
{ulysses,rabbit}!smbhenryb@microsoft.UUCP (Henry Burgess) (07/22/83)
Wizards: Sory about this note, I just can't get mail to smb. smb: We are happy users of 3com hardware and software. I am very interested in your 4.1c/4.2 problems, I will have your problems as soon as we switch over. I can't get mail thru to you. Please send mail via a route I can respond to. Henry Burgess decvax!microsoft!henryb (206)828-8080
drockwel%bbn-vax@sri-unix.UUCP (07/25/83)
From: Dennis Rockwell <drockwel@bbn-vax> A warning about compatibility with 4.1c/4.2: the TCP regularly overruns the receive window (usually by a factor of two). This causes extremely poor performance when the 4.2 side is sending to a non-4.2 system, as it waits for its retransmit timer to go off to resend the out-of-window data that got dropped.
cak@purdue.ARPA@sri-unix.UUCP (07/27/83)
From: Christopher A Kent <cak@purdue.ARPA> We at DEC Western Research Lab are using 3Com boards with DEC taps and 4.1c; the stuff seems fairly robust (except for one board that came in with the wrong address in it's prom -- that took a week to find). The people at 3Com have been helpful, if slow in turnaround on repairs. I don't know anything about their software. The fact that you can't send to yourself is a lose, but Berkeley kludges around it by pushing that stuff through the software loopback. Even though the boards are reputed to be the slowest Unibus Ethernet boards on the market, we're still wire limited by TCP. I haven't tried to measure how much of a load all the copying is. Cheers, chris