But if there is no other traffic, why would you need deep packet buffers to beat records? (It doesn't say if this was on a production network with other traffic or not, so I don't know)

I wouldn't suspect its on a production network at all. :slight_smile:

we might disagree wrt the definition of "production" :slight_smile:

shows the path between caltech and cern transiting

- a GSR in CENIC's LAX POP -- this is a production router for many
  higher ed institutions in california, including all university
  of california campuses, stanford, USC, caltech, and the california
  state university system.

- abilene, the internet2 backbone, which is a production network for
  many higher ed institutions in the US. the abilene "weather map" image
  shown at the URL above shows that indeed there was other traffic
  on some of the links transited.


And likewise so for the single-stream record holder, Sunet.
The only non-average-production links on the path are the connections
between Sunet and Sprint (a separate peering for this, between two
otherwise production routers) and the end tails; not all PC machines on
LuleΓ₯ University have 10GE connections to interfaces in the core of Sunet.