Havard.Eidnes@runit.sintef.no  wrote:
 > If I'm not much mistaken, the SDH STM-1 (which for these purposes
 > should be the same as SONET OC-3) section overhead removes 9/270 of
 > the bandwidth and the path overhead consumes another 1/270, giving you
 > 155.520Mbit/s * 260/270 which comes out to 149.760Mbit/s which can be
 > used for ATM cells.  If the above mentioned value was observed with
 > ttcp you'd have to additionaly take into account the ATM cell header
 > overhead (which comes out to approx. 10%), possible AAL5 encapsulation
 > overhead and the overhead possibly caused by the last cell in the IP
 > packet not being completely filled with "user" data.

well, I can't argue with the math.  Maybe there's something wrong with
'ttcp'?  I did observe it and even pasted it into a conversation I was
having with a number of folks at the time, including Jason and Charles
as well as others whom I can't remember... 

On a day to day basis however, I only use at most, 10mbit of the link 
since that's the size of our active PVC.