Re: [tcpm] I-D Action: draft-ietf-tcpm-hystartplusplus-01.txt

[Jeremy Harris <[e-mail]>]

On 06/01/2021 18:29, [e-mail] wrote:
> 	Filename        : draft-ietf-tcpm-hystartplusplus-01.txt

> Abstract:
>      can cause overshotting of the ideal send rate

Perhaps a typo for "overshooting"?   Suggest just "overshoot".


Section 1 para 2:
>    [HyStart].  After the HyStart delay algorithm finds an exit point,
>    LSS is used in conjunction with congestion avoidance for further

It's unfortunate that "LSS" was expanded in the Abstract.  Readers who
dive straight into the content, expecting it to be complete, will be
confused.

> LSS is used in conjunction with congestion avoidance for further
>    congestion window increases until the first packet loss is detected

Which congestion-avoidance algorithms are expected to operate in
conjunction with LSS?



Section 4.2:
- appears to be describing the delay-increase detection, which has
already been incorporated by reference to the original HyStart.
Avoid duplication.

- also describes the LSS algorithm, which was incorporated by ref.
from RFC 3742.

Section 4.3
- Perhaps LOW_CWD should be defined as a factor of IW, given the
discussion of the effects of altering it?

- MIN,MAX_RTT_THRESH could be renamed MIN,MAX_RTT_DELTA_THRESH.
I at first thought that 16ms was unusable for a sat-link, until I
realised it was a bound on the *increase* in RTT.

-- 
Cheers,
   Jeremy