[c-nsp] IGX UXM bandwidth weirdness

[R. Raja]

Thanks. I will attempt that. Is that value pretty much a trial-and-error
change?

Suresh


On 3/21/06, Mark Rogaski <wendigo at pobox.com > wrote:
>
> An entity claiming to be R. Raja ( r.raja05 at gmail.com) wrote:
> :
> : We are trying to switch the E1 trunk to a UXM card.  When we do this,
> the
> : bandwidth usage by a 24 channel UVM connection remains the same value ie
> 700
> : packets/sec. This is unexpected since we expect each UVM connections
> usage
> : to drop by about 50% (to about 350cps) since a ATM cell is 53 bytes and
> FP
> : cell is 20 bytes.
> :
> : Is this value configurable somewhere?
> :
>
> StrataCom fastpackets are actually 24 bytes, still small enough that you
> can pack the payloads down to get 50% reduction in the cell rate.  The
> problem is that the trunk card is sending ATM cells as quickly as possible
> and isn't waiting to fill up the payload.  You can adjust this by changing
>
> the combining delay with the cnfcmb command.
>
> http://www.cisco.com/en/US/products/hw/switches/ps988/products_tech_note09186a00801da57f.shtml
>
>
> Keep in mind that this is a space-for-time efficiency trade.
>
> Mark
>
> --
> []                           |
> [] Mark Rogaski              |          Politics is the entertainment
> [] wendigo at pobox.com         |          branch of industry.
> [] mrogaski at cpan.org         |                   -- Frank Zappa
> []                           |
>
>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.1 (GNU/Linux)
>
> iD8DBQFEIMzRe3p4SwVZZdwRApILAJ0Q1J+bJSzASw21GfqypDSaNhyTJwCfYr5J
> lTQ8P+l3B4lXHFjCJ7Dn+X4=
> =Zrnt
> -----END PGP SIGNATURE-----
>
>
>