I'd say that in the long run (it was like this when I deployed ethernet dslams 6+ years ago) we're headed for the same kind of behaviour in the ISP end as ETTH switches has today, ie few milliseconds of buffers and 4 strict prio queues (think ETTH but with VDSL2 as PHY instead of 100BASE-T).
So if the ISP->CPE path behaves like this, would it perhaps be nice to have the CPE->ISP direction behave the same way?
-- Mikael Abrahamsson email: [email protected]
