[j-nsp] Slow performance of the KRT queue

[Saku Ytti]

On 5 February 2016 at 19:25, Raphael Mazelier <raph at futomaki.net> wrote:

Hey,

> I have to say, that even if the design problem remain, two minutes isn't
> that bad. In the first day of MX80 with flowing enabling it took age from
> the rib syncing down to the fib (friends report 20min in the worst case).
> Juniper have made enhancement (or fix) in the last junos.

I think the fundamental problem here is that these fixes are
attempting to make the symptoms less pronounced, rather than address
the problem.

I view the problem as desync of software and hardware state, we can
advertise BGP route and attract traffic, even though we have not
actually programmed that state to hardware.
There should be inherent guarantee that what we claim is true in HW. I
can accept that it takes time, that's separate problem.

I'm sure JNPR has the talent needed to fix this, but it might be more
scary fix with wider impact on the infrastructure than management is
ready to sign off.
-- 
  ++ytti