My original question to the sun-nets mailing list was:
> I've been asked to comment on a proposed wiring scheme for a new building that
> NCAR is moving into. One of the items under discussion is how to provide
> ethernet access in every office. NCAR is considering either wiring thin-net
> or twisted pair ethernet from each office to a nearby wiring closet. The
> wiring closet would contain thin-net or twisted pair ethernet fanout boxes
> which would in turn connect to a back-bone network.
>
>
> (Our existing networks are thicknet, but since thicknet transceiver cable is
> difficult to run, it might only be available by special request in our new
> building .)
>
> If anyone has any experience using twisted pair ethernet by itself, or with
> mixed nets of thick and thin net, I'd love to hear from you. I'll summarize
> if anyone shows interest.
Of the responses I received, 15 people favored twisted pair ethernet,
while 4 people preferred thin-net.
For example, Gregory S. Baber (uunet!warlock!gregb) writes:
----------------------------------------------------------------------
"We use Synoptics unshielded twisted pair ethernet stuff here and I
can't sing its praises enough. It's flexible, easy to run and
configure, robust, and has not failed once since we installed it a year
and a half ago."
----------------------------------------------------------------------
At least one person planned to replace existing thin-net runs with
twisted pair.
I'll attempt to summarize the information I collected, quoting freely
from various respondents. (If I've misquoted anyone, I'm sorry, but I
believe I have fairly represented everyone's comments).
Cabletron and Synoptics were the recommended vendors for twisted-pair
enet gear. (One respondent had tried ODS, but wasn't pleased with the
result.)
Claimed advantages of twisted pair ethernet were:
1) Ability to manage the network on a per device basis. Easier fault
isolation as a result. (The Synoptics equipment can give you error
statistics per/device, and can even shut off network access to nodes
that are causing problems on the net).
2) More reliable networks. Machines can be moved without having to
terminate the old connection. (Although AMP has a $50 gadget that
automatically supplies the appropriate termination when you disconnect
a machine). Most cable failures only affect a single node. Twisted
pair is less likely to short than coax (one person reported reliability
problems with thin-net shorting at the BNC connectors)
3) "It's HARD to wire each office with thinwire. Too many tricks are
needed to finesse the problem of zero cable between T-connector and
system ethernet interface."
4) "Much easier to wire a new system, since you don't have to pop
ceiling tiles to find a tap."
5) Since most offices have a few extra twisted pair available, it is
much easier to use this for ethernet, rather than trying to string new
cable, and stay within the fire codes, which only allow Teflon jacketed
cables ($$$) in open ceiling space. Twisted pair also supports RS-232,
appletalk, and even video, with the right equipment.
Claimed advantages of thin-net were:
1) It's possible to daisy chain additional workstations in the same
office for lower cost. (However, there are twisted pair to thin-wire
converters available from Cabletron)
2) Repeaters for thin-net cost less than twisted pair, although coax is
clearly more expensive than twisted pair wiring. This is probably not a
significant issue if you already networking concentrators together via fiber.
Other useful information:
Wiring hints:
The general consensus was to get as many twisted pairs into the office
as practical. One person suggested getting a cable of 25 twisted pairs
from the wiring closet (or computer room) to each office, totally
separate from any phone wiring. The ideal seems to be to have a run of
thin-net coax, twisted pair, and several fibers for future use. (Some
people used shielded wiring for RS-232, others didn't. Apparently,
unshielded twisted pair is just fine for twisted pair enet).
Various people suggested keeping wiring runs as short as possible,
avoiding any punchdown blocks between the office and the twisted pair
fanout unit. (Each punchdown acts like ~30feet of extra cable).
Concentrators:
The twisted pair "concentrators" (intelligent hubs) can connect to each
other via fiber, thicknet, thin-net and twisted pair. Some of the
concentrator vendors will be supporting direct connection of SAS-Class
B FDDI nodes in the future. The intelligent hubs do support SNMP for
network management.
Some very rough pricing:
Thicknet to twisted pair transceivers are on the order of $150-200,
with concentrators costing $160-190/node, depending on the scale of the
network. Prices are list from Synoptics (408-441-8023). GSA pricing is
available. I haven't talked to Cabletron about pricing yet.
Oh, by the way, NCAR is proposing to run fiber to each office for
future use.
I asked for help from sun-nets, and I certainly got it!
Thank you, one and all! This was very helpful information, since I had
no first hand experience with twisted pair ethernet.
(If anyone wants, I'll send you a file of all the responses, if you
want to wade through all of them).
Joe VanAndel Internet:vanandel@ncar.ucar.edu
NCAR - RSG
P.O Box 3000 Fax: 303-497-2044
Boulder, CO 80307-3000 Voice: 303-497-2071
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:05:57 CDT