Cell Relay Archive

["Beckerhof, Onno" <Onno.Beckerhof@alcatel.nl>]

Hi Oliver,

Most important to find out is WHERE you detect those errors. Are they
detected on the ATM Switch RX, on the ATM card RX or on the server RX ?
Are you using SVC's or PVC's ? 

Based on the errors as described the mayor problem is the invalid HEC on
the ATM cells. For some reason the device where you detected those
errors found invalid HEC values on the incoming cells. Performance of
the feeding device (the one sending those cells with invalid HEC) could
well be the source of the problem. If you detected the errors on the
Fore switch port connect to the ATM card, the ATM card is doing
something wrong here. On the other hand I understand that when the
throughput is lower then 25 Mbps NO invalid HEC values are detected (eg
everything is okidoki). This could be evidence that this card could be
designed for 25 Mbps operation and is just offering a physical 155 Mbps
interface.

On the other hand you could check some other points:
*When using PVC's between ATM card and Fore switch. Check PCR and MCR
rate (assuming AAL-5 is used).
*When using SVC's between ATM card and Fore switch. Check if the
signalling parameters are correct (for example if correct UNI version is
used etc.) During call establishment also bandwidth related information
elements are exchanged between switch and cpe. 
*Do both devices support Resource Management cells (assuming AAL-5 is
used)? Normally AAL-5 will be used for transport of TCP/IP. Using RM
cells the throughput can be trottled by the switch to avoid congestion
behind the switch. It could be that the ATM card is not supporting the
RM cell mechanism and therefor is sending to much cells to the switch.

Although the tree points above are always worth checking  I do not think
they will solve your invalid HEC values.
 
Cheers,
 
Onno			onno.beckerhof@alcatel.nl