Re: [tcpm] [tsvwg] inband signaling

[Brian E Carpenter <[e-mail]>]

Lin,

On 16/03/2018 13:31, Lin Han wrote:
...
> [LH] Same as using L4S, AQM still try to provide the faired resource allocation to all TCP flows by the rule of max-min fairness. This works well if the link capacity is big enough and all TCP flows will get what its application want. But when the link is congested, we cannot give some critical application the expected bandwidth, each flow will get the equal instantaneous bandwidth determined the total number of flows. To overcome this difficulty caused by TCP fairness, we have to reserve the resource for some application and exclude them from the resource sharing with others if the resource is not enough (treat those flow differently as others, i.e, different queuing/scheduler, etc).

That's pretty much exactly the philosophy behind diffserv, with "flow"
defined as "all packets marked with the same diffserv code point".

The problem is scaling, and always has been. Diffserv sets a hard
scaling limit at 64 classes. With modern technology, what value between
64 and infinity is a reasonable limit? (The discussion on scalability
in the draft is rather brief.)

Regards
    Brian