| hi surrender
The FDDI specs are pretty clear about the characteristics of the
various pmds. So if the customer absolutely wants FDDI over this fiber
I think he is out of luck. However if the customer wants to run
multi-protocol data at ~100 megabits per second he has some options.
He needs two things: a carrier for his fiber optic, and a data-link to
for his routers etc. to access this carrier.
The best approach for the future for this customer might be a private
Sonet (SDH) Lightwave carrier system. On this carrier there is a
standard to run ATM which can connect his data processing equipment
(and voice and Video). Note that Digital will be soon making a big ATM
announcement and there is a lot of information about ATM on
SCHOOL::ATM$PUBLIC:
(you might want to also mail Fred Goldstein about a proposed FDDI over
SONET mapping but why bother when ATM will be so standard)
While this approach is pretty new so your customer might expect a
number of equipement availability and interoperability problems.
If the customer needs immediate access to this net he might want to get a
proprietary transmission systems (Maybe with an upgrade path to SONET
SDH) that provides T1/T3 framing for these links. Then he can use
standard muxs, routers and PBX over the T1/T3 links..
Alcatel, Fujitsu, and NEC are all good vendors of transmission
equipment in the Far East.
|
| This sounds like the same question I just answered in Mail, so I might as
well post my comments here...
First of all, on the question of the distance limits:
The station distance for SMF isn't a fixed limit. Instead, the limit is on
the link loss (or, equivalently, the received power). I think 40 km is
mentioned mostly because a reasonable SMF installation shouldn't have any
problem meeting the loss limits at that distance. If you need 60 km, it
may well be that this can still be done; the way to find out is to do
the link power budget calculations for the particular choice of fiber
cable used, the number and quality of connections and splices to be used,
etc.
As for the total ring size, we've done other bids for rings somewhat in excess
of the standard 200 km limit. The actual critical parameter is the total
ring latency (D_Max). The ring circumference is one factor contributing
to that; station latency is the other. The 200 km rule allows for a lot
of stations. If you have a topology with a small station count, you can
have more fiber without exceeding D_Max. Again, the way to find out is to
do the calculations.
If analysis shows that this all works out, it would be wise to make sure
the people who have to support the customer buy into it. It would be
embarassing to build a working ring, only to find out that local people
don't believe in it because it doesn't go by the book as they understand
it. If you run into trouble on that front, we should be able to help, because
that should only involve explaining the background behind the "cookbook"
rules and why the topology in question is valid and supportable.
Second, on the subject of repeaters:
Be careful with that. If "laser repeater" means something that simply
amplifies the signal (as opposed to something that looks roughly like
a concentrator with only two ports) it may not work properly. There is
an analogous case: some people use outboard boxes by various vendors that
convert MMF to SMF; those often do not work well and in particular not
at long distances. The problem is that they introduce distortion in the
signal, and timing problems. Just because there is enough light doens't
mean that the signal is acceptable.
It's best to avoid devices of that kind. Devices that are actual FDDI
stations (what I mentioned above as "concentrators with two ports") are
of course acceptable. Failing that, if you do use such things, be sure
to check out the resulting system THOROUGHLY.
And finally...
The configuration rules for FDDI are just like those for Ethernet: they
are designed to let you build networks that will work. In both cases,
it is possible to build networks that do NOT meet the rules and yet appear
to work. That may be for two reasons:
1. The configuration is in fact legal, but uses a combination of choices
not described by the configuration rules because the rules do not list
every possibility, only those that are reasonably easy to describe and
understand.
2. The configuration is NOT legal and is NOT reliable, but you've been lucky
so far.
There are a lot of case #2 around, unfortunately. Those are dangerous, because
while you may like having won the bid, you won't like having to keep losing
money on the support contract!
For example, if some competitor is reckless enough to bid a 6 km MMF link,
I'd let them go ahead and fall all over themselves.
However, the case you mentioned may well be ok. I don't have the SMF link
budget numbers memorized, but the link losses mentioned sound like they may
work. Check the detailed specs. And the total ring size is probably not
a problem either. In any case: do the detailed analysis (or get help doing
it if you need it) and go ahead with the bid if the analysis says it's ok.
If it says that it's not ok, stay away!
paul
|