[c-nsp] ASR9k Bundle QoS in 6.0.1

[Brian Turnbow]

> 
> On 12 May 2016 at 15:06, Robert Williams <Robert at custodiandc.com> wrote:
> > Although thinking about it, can't the central arbiter simply treat it
as one
> virtual queue and distribute the 'tokens' to multiple NPs in a
round-robin style?
> Would it be that significantly different from sending multiple separate
streams
> to separate NPs as in the current model (when used for egress queueing
multiple
> ingress ports to a single egress port)?
> 
> There isn't any central arbiter which is aware of all request/grants.
> And even if there was, it has no visibility on port's queues. So that
wouldn't
> work. Only thing I can think of, is counter synchronisation which does
not seem
> practical at all.
> Maybe if each NPU would have another interface towards some shared
> memory, I guess that might be possible in theory, but unsure how much it
would
> impact BOM/thermal/pincount and if anyone would be willing to pay the
> premium.
> 

There is a similar yet different feature in  BNG, shared policy instances,
that permits sharing of a policy between subinterfaces where 

All sub-interfaces that use a common shared policy instance and are
configured on a Link Aggregation Control Protocol (LAG) bundle must be
load-balanced to the same member link.

It was left up to you to do so,  maybe they found a way balance  traffic
towards only one of the interfaces when the policy is present.
Should be easier than sharing queues etc and could make things lots of fun
for the poor folks that apply a shaper and subsequently saturate a link...


Brian