Re: [LOOPS] draft-li-tsvwg-loops-problem-opportunities and use-cases.

[Carsten Bormann <[e-mail]>]

Hi Jeremy,


> On 2020-07-30, at 15:50, Jeremy Harris <[e-mail]> wrote:
> 
> On 29/07/2020 22:52, Carsten Bormann wrote:
>>> (2) I will argue there is not a satellite-specific use case really, there is a use case for any reliable long-haul path (large RTT) that is concatenated with another segment with a lossy segment with lower RTT.
>> 
>> Yes!
>> (For a definition of “reliable”.)
> 
> Seems you'd do well, for that scenario, to have a separate LOOPS-like
> reliability segment for the lower-RTT but lossy segment - to better
> respond given the segment RTT being lower than the full path RTT?

Yes.

> But we're regressing to the 1950's era of separate link-reliability
> mechanisms, and a sequence of such links.  Something the e2e principle
> discarded very successfully.  Are we really better off, provably?

I do not think there is a conflict with the e2e argument here (*).
LOOPS is not replacing TCP/QUIC/etc., it is assisting the e2e protocol.
All links that do not have ridiculous signal/noise budgets have some form of error recovery; that may just not very visible or well known (e.g., Trellis Coding for Gigabit Ethernet).  In theory, that would always be working well and be enough so the rest can be done in the e2e protocol.  In practice, there is a difference between theory and practice.

Grüße, Carsten

(*) Please see Section 2.3 of https://dl.acm.org/doi/10.1145/357401.357402 if you need an appeal to the original reference.