| Yes, absolutely.
There are two quite separate considerations you need to deal with:
1. Requirements on the individual connections. A connection is the pair of
links between two ports. In other words, the duplex fiber, or thinwire
coax, or whatever, between a concentrator port and an adapter is a
connection.
2. Requirements on the ring as a whole.
You can choose the type of medium for each connection totally independently
from every other connection. There may be some packaging limitations -- for
example, the thinwire ports on the concentrator come 6 at a time on a card.
But there are NO rules anywhere that require a particular medium type just
because some particular medium was used elsewhere in the ring. You do need
to use the same for both links of the connection. In other words, if you
use multimode fiber from the M port transmitter to the workstation receiver,
the fiber going from that same workstation's transmitter to the M port receiver
must also be multimode. That shouldn't be a surprise -- or a limitation.
If you look in the Network Buyer's Guide, you'll see concentrator line cards
that have singlemode A, multimode B ports, or vice versa. You would use those
if one connection needs to go a long way but the other does not.
For each medium you might use in a connection there are limitations specific
to that medium for distance etc. For example, multimode connections are
limited to 2 km; singlemode ones to about 40 km (though the real limiting
factor is attenuation); coax to 100 meters or whatever, and so on. Again, the
rules for each individual connection depend ONLY on what medium is used for
THAT connection.
As for the requirements on the ring as a whole, the main one is that the
total circumference must be such that the ring latency does not exceed D_Max
(1.617 ms). If the circumference is close to 200 km, you need to look into
this carefully; if you also have a high station count you may be over the limit.
(Conversely, if you're REALLY careful you can go somewhat over 200 km so long
as there are only a few stations. If you don't understand precisely how to
calculate ring latency, do not attempt to design such configuration!)
paul
|
| Thanks for the speedy reply!
As for calculating the Ring circumference - ring latency - can you give
me a pointer to a document that discusses this, and what the guidelines
and variables, formula if one exists, to do this? That would be great!
Angelo
|
| Ring latency is in principle a simple matter. Add up all the link lengths
that make up the ring. (Be sure to allow for lengthening caused by wrap!)
Convert length to delay (by dividing by the propagation speed). If you're
using media with significantly different propagation speeds, do the arithmetic
separately for each type and add the results. Now you have the contribution
from the links themselves. Add to that the contribution from the stations;
you'll need to know the latency of each station. One microsecond is a likely
number, but some are lower and some may be higher. Add everything, and you
have the answer!
paul
|