[c-nsp] Is this throughput on ATM DS3 in 3640 normal?

[Lincoln Dale]

Mark Rogaski wrote:

>An entity claiming to be Dan Hollis (goemon at anime.net) wrote:
>: On Wed, 31 Aug 2005, Lincoln Dale wrote:
>: > if you end up above the policy they have set then they don't necessarily
>: > drop what you want - in fact - they may drop one cell in the middle of
>: > an AAL5 stream rendering up to 31 other cells as useless with 1500-byte MTU.
>: 
>: If cell are dropped, it will be visible as AAL5 CRC errors on the 
>: interface.
>: 
>
>Not necessarily.  If EPD (frrame-based discard) is used by the provider
>only intact PDU's will be delivered.
>  
>
all good information.  cisco-nsp certainly remains a forum with a higher 
than average clue quotient.

just to follow-up on my experience(s) with all of this ~10 years ago (i 
have nothing to do with this area inside cisco :) ..

we thought that also.  however we also found that there was no guarantee 
that an ATM switch with congestion wouldn't actually guaranteed to do this.
this was almost 10 years ago - so perhaps things have changed since then 
(vendor N ATM switches) - but ... based on the fact that noone has ever 
bothered to try to find a better way to shoe-horn IP-mix-size packets 
into ATM any more efficiency than is currently done suggests probably 
not ...


cheers,

lincoln.

>http://puck.nether.net/pipermail/cisco-nsp/2005-August/023350.html
>
>Mark
>
>  
>