[c-nsp] do i *need* DFCs on the 6500?

[Rob Shakir]

On Wed, Sep 02, 2009 at 02:42:31PM +0200, Peter Rathlev wrote:
> Agreed, and introducing DFCs can give you some headaches with e.g.
> policers, since each forwarding engine operates independently from the
> others.

I guess here you're referring to the fact that there is no communication of
ingress rates between individual DFCs, or back to the PFC? I posed this question
to Cisco at a CPOC lab, and have also heard the same from TAC. Consider the
following scenario:

Two flows are ingress on a 6500, one on Gi1/0 (where slot 1 has a DFC), and one
on Gi2/0 (where slot 2 also has a DFC). They are egress on a third port Gi3/0,
which also has a DFC. Since the policing is performed by the ingress LC, if
Gi3/0 has a 50Mbps policer, and the flows at Gi1/0 and Gi2/0 are both 40Mbps,
neither card will drop any packets, and hence the egress rate at port Gi3/0 is
80Mbps, despite having a 50Mbps policer configured.

Unfortunately, I don't have the resources to test this - has anyone tried this
scenario, and verified this is _actually_ how things work, rather than
theoretically?

(I think the official answer for this was..."This will be fixed in EARL 8")

Thanks,
Rob

-- 
Rob Shakir                      <rjs at eng.gxn.net>
Network Development Engineer    GX Networks/Vialtus Solutions
ddi: +44208 587 6077            mob: +44797 155 4098
pgp: 0xc07e6deb                 nic-hdl: RJS-RIPE

This email is subject to: http//www.vialtus.com/disclaimer.html