[j-nsp] what’s the story behind MPC5E

[Saku Ytti]

On 21 May 2016 at 16:15, Adam Vitkovsky <Adam.Vitkovsky at gamma.co.uk> wrote:
>> I'm very very dubious if there is business case to plan to congest XL.
>> I consider lookup congestion fault which needs to be fixed.
>>
> Well during DDoS attack the XL will get congested so you better be prepared for that.

I disagree. It will only get congested if you have 0 background or you
have background consisting of only small packets or atypical lookup
cost. You should design/populate ports so that this is non-issue.
If you know you have atypical lookup cost or only small packets, then
you should leave some ports unconnected, so that you can guarantee
there is margin.

In vast majority of networks there is IMIX background of >500B average
size, then even with modest background you have nothing to worry
about, you can't fit more work on the wire than the LX can handle.

And this issue also has nothing to do with XL sharing, even in
unshared situation you have same problem, you may congest XL, it's not
wire rate.

-- 
  ++ytti