Re: [multipathtcp] MPTCP Schedulers

[Vladimir Olteanu <[e-mail]>]

On 2/17/19 12:20 PM, Olivier Bonaventure wrote:
> Vlad,
>> I'm trying to tie up any loose ends in the SOCKS 6 draft. There is some
>> functionality which lets you change the MPTCP scheduler used by the
>> proxy. The available schedulers are basically the ones available in the
>> Linux kernel implementation.
>> (https://tools.ietf.org/html/draft-olteanu-intarea-socks-6-05#page-20)
>>
>> Do you think it would be worthwhile to standardize the MPTCP schedulers?
>> I think an informational or best practice RFC would be useful.
> I think that this would make sense, but we could also encode the
> requested MPTCP scheduler inside an MPTCP option instead of putting it
> in an application level message as you propose in your draft. The
> advantage of using MPTCP options is that the same solution would work
> for any application.
>
>
> Olivier

Hi,

I agree that it is worthwhile doing it at L4. As far as SOCKS is 
concerned, this would also open the door for the client controlling the 
scheduler used not only by the proxy, but also by the remote host.

I'd argue that being able to do it at the application layer is also 
useful, even if your proposed option gets standardized. The scheduler 
option will not fit inside a SYN + MP_CAPABLE + TFO (along with SACK, 
TS, MSS and Wscale) and you'll have to send it 1 RTT later. (In fact, 
not even the TFO option will fit if the cookie is large.) OTOH, there is 
on such pressure oh the SOCKS option space; you've got 16K to play with.

Cheers,

Vlad