Sun-Spots Digest, v6n121
William LeFebvre
Sun-Spots-Request at RICE.EDU
Fri Jun 24 14:05:31 AEST 1988
SUN-SPOTS DIGEST Thursday, 23 June 1988 Volume 6 : Issue 121
Today's Topics:
Looking for feedback on 15-pin Ethernet connectors (9)
DELNI's on Suns
Re: Pointers to performance studies on TCP/IP
Send contributions to: sun-spots at rice.edu
Send subscription add/delete requests to: sun-spots-request at rice.edu
Bitnet readers can subscribe directly with the CMS command:
TELL LISTSERV AT RICE SUBSCRIBE SUNSPOTS My Full Name
Recent backissues are available via anonymous FTP from "titan.rice.edu".
For volume X, issue Y, "get sun-spots/vXnY". They are also accessible
through the archive server: mail the request "send sun-spots vXnY" to
"archive-server at rice.edu" or mail the word "help" to the same address
for more information.
----------------------------------------------------------------------
Date: Mon, 13 Jun 88 11:57:24 PDT
From: dmc at tv.tv.tek.com
Subject: Looking for feedback on 15-pin Ethernet connectors (1)
[[ This note was also posted to the tcp-ip list, and it produced a flood
of responses. I am including that discussion in this issue. --wnl ]]
I am looking for feedback on the quality and reliability of the slide
latch mechanism used on the DB-15 ethernet connectors. I have used these
myself for the past five years, and have discovered a number of problems
with them. Additionally, various nasty remarks about these connectors
have surfaced on the net from time to time.
The Society of Motion Picture and Television Engineers has a standard
known as RP-125 which specifies a parallel digital television interface.
The connector for that interface is a 25 pin D-type (DB-25) with a slide
latch like the ethernet connector. The RP-125 standard is up for
revision, and the Tektronix representative submitted a note from me
outlining my experience and problems with the ethernet slide latch
connector.
Sun Microsystems also has a representative on the RP-125 committee. He
has asked the component engineer for connectors at Sun if there was any
data concerning the mechanical integrity of the slide latch locking
mechanism used in the ethernet connector. Here is the response:
"I am the Component Engineer for connectors at Sun. In regards to your
question concerning the Ethernet DB-15, these are used through out
industry by the millions, in many applications. I am not aware of any
significant problems with mechanical integrity of the lock, or for that
matter with the connector in any aspect. If the connector is properly
installed and handled in a semi-reasonable manner you should not have any
problems. As far as Sun is concerned, usage of D-Sub connectors in
general, that is various sizes and configurations, will probably increase.
When all things are considered it often turns out to be the best way to
go! Hope this helps!!!"
In the face of this informal statement by Sun's component engineer, I now
need data about other people's experience with the DB-15 slide latch
ethernet connector. Help save another standard from the same fate!
Please send your comments email to:
dmc at tv.tv.tek.com
or
tektronix!videovax!dmc
Thank you,
Don Craig
Tektronix Television Systems
------------------------------
Date: 14 Jun 88 15:28:03 GMT
From: phri!roy at nyu.edu (Roy Smith)
Subject: Re: Looking for comments on the 15-pin ethernet connector (2)
dmc at videovax.Tek.COM (Donald M. Craig) writes:
> I am looking for feedback on the quality and reliability of
> the slide latch mechanism used on the DB-15 ethernet connectors.
In a nutshell, they suck! There is no doubt that the single most common
cause of problems in our entire network (19 Suns, a Vax, and maybe a dozen
Macintoshes and PCs) is loose ethernet tranciever cables, particularly on
the backs of Sun-3/50's which provide no mechanical support for the cable
at all. On our rack-mount systems, we support the cables with cable ties
to various convenient supports. On our deskside suns, we've constructed
assorted mechanical strain reliefs. Some of our 3/50's seem to be OK with
just wedging the cable behind a desk but some are a constant cause of
trouble. For the worse ones, we install a support bracked we've designed
which helps a little (it's just a plexiglass bar notched to fit on the
card extractor ears and with cutouts for the various cables and attachment
points for cable ties in the appropriate places).
> "I am the Component Engineer for connectors at Sun. In regards to your
> question concerning the Ethernet DB-15 [...] I am not aware of any
> significant problems with mechanical integrity of the lock, or for that
> matter with the connector in any aspect.
I'm not given to public flamage, but this guy must have his head firmly
wedged in a dark place. If he's not aware of any problems, it because he
hasn't been listening. I've complained loudly about this on the net
before. I've complained to Sun field service. I've complained to Sun
tech support. Clearly those complaints havn't gotten back to the right
people.
The stupid little stamped sheet metal clips are simply not strong enough
to secure a connector with a big fat, heavy, and fairly stiff tranciever
cable on it. As long as the cable is secured so it can't move if
accidentally moved, it's OK. For example, on the tranciever ends, we lash
the cable to the main ethernet trunk cable with 2 (or sometimes 3) wire
ties a few inches away. But on systems which might move a little (like a
deskside Sun on wheels), or in situations where the cable might be
disturbed (like hanging off the back of a desk) forget it.
What was wrong with good-old RS-232-style screws? Or, if they really
wanted a tool-less installation, why not Macintosh-style knurled screws,
or maybe even centronics-style wire bails? We recently got a 3-Com 3C503
ethernet card for an IBM-PC. The connector is a bit different, with screw
holes instead of binding posts. Unfortunately, to use the screw holes you
need a special adaptor bracket which I havn't been able to locate yet (OK,
we just got the thing; I havn't had a chance to look very hard). It looks
like it might be a bit more secure. Our Interlan ethernet board for the
vax has a slight variation on the slide connector which looks like it
might be marginally stronger (it has small extra ridges along the sides),
but I doubt it would still be strong enough if we weren't able to lash the
cable to various places in the vax's rack frame.
I really don't know what the DIX guys had in mind when they designed this
connector. Administering a network is hard enough without having to worry
about which $5 connector is falling out.
Roy Smith, System Administrator
Public Health Research Institute
455 First Avenue, New York, NY 10016
{allegra,philabs,cmcl2,rutgers}!phri!roy -or- phri!roy at uunet.uu.net
------------------------------
Date: 14 Jun 88 17:58:28 GMT
From: ron at TOPAZ.RUTGERS.EDU (Ron Natalie)
Subject: Re: Looking for comments on the 15-pin ethernet connector (3)
Yep, by the way, we've gone and replaced the slide clips with the good old
RS-232 screws and threaded inserts on all the Ethernet cables of
importance.
-Ron
------------------------------
Date: 14 Jun 88 19:54:09 GMT
From: bzs at bu-cs.bu.edu (Barry Shein)
Subject: Re: Looking for comments on the 15-pin ethernet connector (4)
I'll second Roy Smith's gripes about those slide-latch ethernet connectors
to quote Roy:
"In a nutshell, they suck!"
Actually, if there were some negative pressure they might work...
We have around 100 Suns here and other assorted items (this lousy
slide-latch is not unique to Suns by any means, it's ubiquitous) and
they're always falling out, the worst catastrophes occur when they fall
out of centralized server machines, we've been out for hours and hours
while some poor operator tries to figure out what the problem is (they're
learning, we're all learning.)
Worse, much worse, it falls out of MY SUN all the time, like when I swing
my chair around to gulp coffee or some other critical maneuver and brush
the deskside tower, to get it to stay back in I have to re-arch the cable
at a precarious angle so it provides pressure towards the plug.
Now *that's* unacceptable, the public be damned.
I'll also add a vote to the Mac-like knurled screws, that would be the
ticket.
-Barry Shein, Boston University
------------------------------
Date: 14 Jun 88 17:41:50 GMT
From: hp-sdd!ncr-sd!se-sd!cliff at hplabs.hp.com (Cliff Bamford)
Subject: Re: Looking for comments on the 15-pin ethernet connector (5)
The problem is that it's impossibly impractical to give that connector
"reasonable care" in the field. Regardless of type, connectors always seem
to be just-barely-visible and just-barely-out-of-reach. The D-sub requires
good visibility (to tell which way the slider is slid) and good tactile
access to [un]couple. Since these conditions rarely obtain, THE !#@$%&*
CONNECTER DRIVES PEOPLE CRAZY.
Which is one of the reasons you see so many of them torn asunder, at which
point they become worse than useless. Except as mute testimony to the
limits of human patience.
cliff.bamford at sandiego.ncr.com (619)693-5724 {ucsd,cbosgd}!ncr-sd!se-sd!cliff
------------------------------
Date: 14 Jun 88 21:02:30 GMT
From: ati.tis.llnl.gov!bae at tis.llnl.gov (Hwa Jin Bae)
Subject: Re: Looking for comments on the 15-pin ethernet connector (6)
In general, I don't like DIX 15 pin adapter without any secure screws. My
Symmetric 375 comes with 9 pin output which can be interface to regular
transceiver with a 9 pin to 15 pin drop cable. You can use RS-232 like
screw attachment with it, and it's so much nicer than SUN's way of doing
attachment. But then again you can always use "Crazy Glue" or
something... 8-)
Hwa Jin Bae |
Control Data Corp. | (415) 463 - 6865
4234 Hacienda Drive | bae at tis.llnl.gov (Internet)
Pleasanton, CA 94566 | {ames,ihnp4,lll-crg}!lll-tis!bae (UUCP)
------------------------------
Date: 14 Jun 88 20:41:12 GMT
From: kwe at bu-cs.bu.edu (kwe at bu-it.bu.edu (Kent W. England))
Subject: Re: Looking for comments on the 15-pin ethernet connector (7)
In article <23340 at bu-cs.BU.EDU> bzs at bu-cs.BU.EDU (Barry Shein) writes:
>
>I'll also add a vote to the Mac-like knurled screws, that would be the
>ticket.
That would just about do it, but there are some little touches that the
hardware guys could do if they just thought about how the equipment gets
used. Hardware engineers--you listening??
Redesign the host side of the connection for better strain relief. We can
handle the xcvr side with ty wraps alright.
First, use a 90 degree angle on the cable into the connector. Then
provide a ty wrap mount an inch or so away from the connector mounting
point so we can tie the cable down and provide strain relief. Barry will
also be able to shove his deskside closer to the wall and that will help.
If you can't angle the cable put a standoff bar somewhere on the back of
the box so that when the unit IS shoved into the wall it won't crush the
cable/connector and pop it off. Ideally, put a strain relief point on the
standoff bar so we can tie the cable down. (Smart guys will do both.)
This would help a great deal even without a major session of 802.3 devoted
to rehashing this.
Kent England, also BU
------------------------------
Date: Tue, 14 Jun 88 23:02:31 pdt
From: Robert Michaels <robert at atom.hpl.hp.com>
Subject: Re: Looking for comments on the 15-pin ethernet connector (8)
Telephone: +1(415)857-3403 (PST/PDT UT+8)
> The stupid little stamped sheet metal clips are simply not strong
>enough to secure a connector with a big fat, heavy, and fairly stiff
>tranciever cable on it...
>What was wrong with good-old RS-232-style screws?
To me the problem is that the little clips are "standard". I would be very
surprised if a vendor will have enough courage to promote something more
effective. To alleviate your problem you could try using a lighter weight
transceiver cable. For us most transceivers are within 5 meters of the
host. For short runs you don't need these really big fat cables. I think
several vendors build these short lightweight transceiver cables ( yes
even HP does).
- Robert Michaels
HP Labs
------------------------------
Date: 14 Jun 88 23:13:34 GMT
From: cs.utexas.edu!execu!dewey at husc6.harvard.edu (dewey henize)
Subject: Re: Looking for comments on the 15-pin ethernet connector (9)
Here at Execucom we have a slightly different term for this 'device'. I
won't put it out on the net, however, since this probably isn't the place
to put adjectives regarding personal ancestory or public sexual habits.
Suffice it to say that if the person who designed the slide latch
connector were to come here and visit, his/her employer should only buy a
one-way ticket - we'd be able to ship what remains there were back in a
shoebox!
Seriously, those things are a joke. When each and every part is made to
exact and perfect spec, they seem to work ok (at least until you touch
them in some fashion). The ones we got with at least 3 of our Suns and
the cables that we also got from Sun seem to be made to different specs
entirely. We had to do a good bit of reverse engineering to get them to
handle the normal vibration of someone merely walking in the office
(concrete floors).
Junk em. Use screws.
Please, since you are protecting this 'Component Engineer', help him/her/it
:-) become aware of 'significant problems'.
execu!dewey Dewey Henize @ Execucom Systems Corp 512/346-3008
------------------------------
Date: Tue, 14 Jun 88 16:21:20 edt
From: sdrc!crken at uunet.uu.net (Ken Shackelford)
Subject: DELNI's on Suns
>From: lrj at helios.tn.cornell.edu
>What problems are there with DELNI's?
I believe that some of the new systems that have both thin and thickwire
ethernet ports have som kind of circuitry for detecting which is attached.
I think it works by trying to detect a power draw from the the
transceiver. If it detects this draw (a certain amperage range), it
attempts to use the thick port. Otherwise, it uses the thinwire. The
problem occurs when you attach one of these systems to a DELNI. The DELNI
is self-powered, and as such, draws no power from your machine. The
detect circuitry finds no draw so you have no ethernet. On the machine we
have, I was able to find a jumper in the hardware documentation that
turned off the thinwire, and hence, the detect circuitry. Once I did that
everything worked fine.
Ken Shackelford (The Wizard of Odd) UUCP: uunet!sdrc!crken
------------------------------
Date: 14 Jun 1988 23:41-EDT
From: KSEO at G.BBN.COM
Subject: Re: Pointers to performance studies on TCP/IP
The SATNET (Atlantic Packet Satellite Network) Measurement Taskforce just
completed some work analyzing TCP/IP performance over the SATNET. As part
of this effort, we developed an instrumented TCP/IP implementation and a
separate IP measurement tool. For additional information:
a) Our experimental methodology, tools and results are
described in a paper to be presented at SIGCOMM 88
in August. Send me your paper address if you'd like a
copy.
b) There will be 3 additional writeups describing the
taskforce's work in more detail. These will probably
be done by the end of the summer. Let me know if you'd
like copies when they're complete.
c) For specific details about the TCP and IP tools, please
contact:
TCP tool: Jon Crowcroft (jon at cs.ucl.ac.uk)
IP tool: Paal Spilling (paal at tor.nta.no)
Let me know if you have further questions,
Karen
------------------------------
End of SUN-Spots Digest
***********************
More information about the Comp.sys.sun
mailing list