[c-nsp] ASR9k Bundle QoS in 6.0.1

[Saku Ytti]

On 8 June 2016 at 14:14, Adam Vitkovsky <Adam.Vitkovsky at gamma.co.uk> wrote:

>> But is that really so? Why can't we do this with 100% separated fabrics
>> with
>> distributed arbiters? The linecard could resend fabric request if it does
>> not
>> receive reply in n microseconds, to handle failover with no loss?
>>
> We are actually talking nanoseconds here.

The chassis delay for non-congested packets are in several
microseonds, perhaps like 10us or bit less. For non-fabric, single NPU
boxes can be under 5us. But how is this 10us budget distributed
amongst lookup, fabric, copying etc. No clue.

And paying rerequest tax during failure does not seem impossible tax.

-- 
  ++ytti