Cell Relay Archive

[fgoldstein@bbn.com (Fred R. Goldstein)]

In article <3ks4f0$roh@GRAPEVINE.LCS.MIT.EDU> mikec@ginger.lcs.mit.edu (Mike Ciholas) writes:
>At the OC3 bit rate, 155 megabits per second, how many bits/sec are
>user data?

>Assuming the 155 mbs number is the serial wire encoding rate, and with
>8B/10B encoding, that leaves 124 mbs for cells, then assuming 48 out
>of 53 bytes are user data, that leaves about 112 mbs.  Is this
>reasoning correct?  Are cells transported in SONET frames (of what
>size and what header overhead for that?)?

There is no 8/10 coding; that's only used in TAXI and some other LAN-style
interfaces.  Begin with STM-1's 270 columns, subtract 9 for fixed overhead
and 1 for floating overhead.  Then just stuff the cells into the remaining 
payload one after the other.  I think it's 149.98 Mbps left.

>I presume that a "155 mbs" ATM interface has two 155 mbs streams, one
>in and one out simultaneously.  Is this true?

Yes, it's full duplex.

>Lastly, where can I find information on the physical plant model for
>ATM?  That is, what kind of wiring/fiber, what do the hubs/switches
>do, etc.

ATM Forum specs show some, ITU some.  There are many options.
HP Idacom puts out a nice chart (they sell ATM protocol analyzers).
Several textbooks cover the subject and are a place to begin.