RE: [nsp] ATM PA input errors.

From: Martin, Christian (CMartin@mercury.balink.com)
Date: Thu Jun 01 2000 - 00:46:38 EDT


 
>
> On Wed, May 31, 2000 at 12:55:13AM -0500, nicholas harteau wrote:
> > Can someone point me towards exactly what 'abort' implies?
> >

This usually refers to AAL5 PDU's which failed a CRC check due to cell loss.
Recall that the CPCS trailer contains a 32 bit CRC. If a cell is dropped by
the UPC function in an ATM switch, the whole PDU is lost. Check with your
carrier to determine if an intelligent packet drop policy can be
implemented. Partial Packet Discard works well in speeding TCP recovery
from loss, thus helping to smooth rate oscillation amplitude under load.
Oh, and the aborts will go away.

.chris

PS

Note that in some 11.1 trains and perhaps in the case of AIPs, CPCS PDU CRC
failures were shown as CRC errors. I suppose people were confusing this
with HEC errors, so Cisco attempted to clarify the diags. Looks like they
missed the mark....

>
> In the ever helpful words of CISCO:
> Number of packets whose receipt was aborted.
>
> I'll have you know I searched cco.cisco.com for 30 minutes
> looking for that.
> :)
>
> --
> $Id: .sig,v 1.24 2000/04/15 17:58:41 noise Exp $
> read this before sending me mail:
> http://rritzman.home.mindspring.com/why.htm
>



This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:13 EDT