If theres only a few HEC's then I wouldnt be too concerned... we'd need to see full stats to get a more overall picture.
>> there is no information about what number of multiple bit errors, i.e. uncorrectable, there are
HEC is a basic type of CRC error check which is performed on the ATM cell. HEC is capable of detecting multiple bit errors, but can only correct errors where a single bit is missing.
The figures usually reported by the router are multiple bit errors that have been detected but unable to be reconstructed from the (simple) checksum. Therefore any HECs recorded would in theory be uncorrected multiple bit errors.
Its similar to CRC count in that an error has been detected but information has been re-requested.
A lot of multiple bit errors need the likes of FEC/Interleaving. The FEC count is one that usually shows a corrected count (rather than uncorrected).. which is why we are often unconcerned about FECs
If you were getting a lot of HECs and becoming a problem, I would also (normally) expect there to be a fair amount of CRC errors. Some routers may display CRC errors as the term RS Uncorrectable Errors.
You dont mention the number of errors recorded against HEC.. but something worth mentioning is that when on maxdsl and a rock steady line is that I would _always_ rack up a minimum of 8/16 HECs during the sync up phase and this number would stay like that for ages perhaps only slightly increasing if I got any CRC's.