RE: ATM vs. DS3

We monitor both the octet output of the cisco, and the received cells
on our ATM switch. One then would divide the cisco octet value by the
cell value * 53. This would result in a percentage value that would
indicate your "efficiency".

Efficiency (%) = octet rate / (cell rate * 53)

This assumes that you can measure both parameters.

Chris A. Icide
Sr. Engineer
Nap.Net, L.L.C.

We go direct Cisco <-> Cisco DS3, no switch. Any variables to measure on the
Cisco, assuming you only have access to 1 side?


Cisco doesn't have a lot of ATM stats avaiable yet. We have been beating
the ATM developers to try to get better stats.. but no luck yet. We have
had a lot of problems the way the SAR chip accounts for VBR traffic.. and
they don't even do CBR. The output/input counters on the interfaces are
not valid ways of measuring traffic either, they don't take ATM Traffic
shaping into account. This is because the interface counters are
calculated before the actual SAR chip segments/reassembles the ATM Cells.
This can be proven by using an ATM traffic generator and flooding the
interface... According to the interface stats I was pushing over 200 mb
through and OC-3. I wish!!

So, in a nutshell, there isn't a real good way to measure
performance/effieciency on a Cisco.. yet.