🚀 go-pugleaf

In article <1992Nov21.162954.17036@umr.edu>, rfranken@mcs213d.cs.umr.edu (Richard Brett Frankenberger) writes:
> In article <1992Nov20.235027.16623@nntpd.lkg.dec.com> mitton@dave.lkg.dec.com (Dave Mitton) writes:
> >	In fact it is trivial to break a token ring totally in many ways:
> >- cut the trunk(as above),
>    Yes, but the trunks don't run to every workstation.  I would have to get to
> the MAUs to do it.  The single ethernet cable runs to every workstation, and if
> it gets damaged, there goes the whole net.
> 
> >- put an adapter cable (no loop on end) in a RI/RO port
> >- crosswire RI/RO ports (took me a while once on this one), 
> 
> Both true.
> I would imagine an ethernet can be brought down by a miswired cable also.
>

Not really. For 10BaseT, the hub can isolate an individual station that
is mis-wired, so that the network is not brought down. In a 10Base5
drop cable, it is not possible to do any miswiring that will cause
network outage. A destroyed transceiver is the worst that could
occur, which just brings the stations down. 802.3 specifications even call
out the fault-tolerance requirements, which include putting 16 volts
on the transmit, receive and collision pairs (mis-wiring the transceiver
power).

As to the coax trunk in either 10Base5 or 10Base2 I defy you to miswire
a coaxial cable and connect the shield to the center pin of the connector
and the center conductor to the connector shell (even intentionally!)
 
> 
> >- always a favorite: insert a station of the wrong speed
> >	(station alternately causes the ring to beacon and drop out
> >	but keeps re-inserting because it's local lobe test works)
> 
> Not true.  The card is supposed to give up after a set number of tries at
> initial insertion.  (It is NOT supposed to keep trying - and MOST do
> not.  HP JetPorts are an exception (I found out the hard way...))
> With workstations; however, the ring beacons for a moment a few times, and then
> the workstation says ERROR INSERTING and stops.
>  
> Note that EtherNet would likely have a similar problem if it supported multiple
> speeds.

Not really. The problem in TR occurs because of passive MAUs. An active
hub can keep a wrong-speed station from joining the ring (many modern
hubs do exactly this). Ethernet hubs are by definition active, and can
avoid this problem.

BTW, if you are allowing for software fixes for fundamentally hardware/
architectural problems (your example: a station should give up tryinbg
to join a ring after so many tries), then either NO network has any problems
(everything can be fixed with sophisticated-enough software), or you
have to deal with the problem that the station is BROKEN, and the
software won't/can't fix anything for you (if it wasn't broken, there wouldn't
be a fault to correct, now).

> 
> So token-ring rules are a bit more complicated.  I never said Token-Ring was
> easier to install.  It isn't (especially when compared to thin-net). 
> True - token-ring at the trunk level isn't much better than Ethernet; but 
> Since ether net at the trunk level is the same as ethernet at the drop level,
> but token-ring is much more robust at the drop level ...
>  

This is only true for 10Base2, which is an OPTIMIZATION of Ethernet for
small workgroups. If you want full Ethernet robustivity, you can
use either 10Base5 or 10BaseT. If you want very cheap, easy to
install, "plug-and-play" Ethernet, you can use 10Base2 with its
increased susceptibility to workgroup LAN failure. The point is
you have a choice. Token Ring requires that you pay more for stuff
whether you want it or not.


-- 
Rich Seifert                    Networks and Communications Consulting
seifert@netcom.com              (408) 996-0922
                                (408) 996-2860 FAX
"... specialists in Local Area Networks and Data Communications systems"