Re: [AVTCORE] Framemarking in video packets

[[e-mail]]

Harald Alvestrand <[e-mail]> writes:
> This only works if the header extension is unencrypted.

Quite true.

It's also possible that it's not worth the trouble.  The full
framemarking might reveal that the video stream is separated into 5
layers with prioritization between them, but that no router would ever
implement complex enough queuing to want to know that.  In my lifetime,
a lot of difficult prioritization problems have been solved by just
provisioning more bandwidth.

Indeed, there's an amusing version of this from my own history.
Traditionally, cloud service providers have the aggregate bandwidth of their
rentable hosts well above capacity of their interconnect network.  This
leads to "data center networking", various tricks to deal with the
situation.

I went to work for a cloud service provider, and part of the reason was
to learn "data center networking".  It turns out that the provider just
fully provisioned the interconnect network so they didn't have to worry
about all that.  Then they billed it as a feature of the service.

Dale