Re: [multipathtcp] Number of DSS to store

[Anumita Biswas <[e-mail]>]

Olivier,

On Aug 27, 2012, at 11:57 AM, Olivier Bonaventure <[e-mail]> wrote:

> Anumita,
> 
>> Continuing on the topic of SACK and on Duplicate ACKs in general; 
>> 
>> Seeking clarification on this statement in the draft:
>> 
>> This changes
>>   the semantics of a duplicate ACK, these are usually only sent as a
>>   signal of a lost segment [9] in regular TCP.  Therefore, an MPTCP
>>   implementation receiving a duplicate ACK which contains an MPTCP
>>   option MUST NOT treat it as a signal of congestion.  Additionally, an
>>   MPTCP implementation SHOULD NOT send more than two duplicate ACKs in
>>   a row for signaling purposes, so as to ensure no middleboxes
>>   misinterpret this as a sign of congestion.
> 
> In this paragraph, "signaling purposes" means a duplicate ACK that is
> only used to convey signalling information, ie. add address or remove
> address.
> 
> During the data transfert, a duplicate ACK can contain the following
> options :
> 
> - DSS
> - MP_PRIO
> - ADD_ADDR
> - REMOVE_ADDR
> 
> A packet should only contain one DSS option but an MPTCP sender could
> need to send several MP_PRIO, ADD_ADDR and REMOVE_ADDR. The paragraph
> above was written to convey the idea that a duplicate ACK containing
> these options should not trigger a congestion response. It seems that it
> is a bit vague.
> 
> Would the following be clearer ?
> 
> "This changes the semantics of a duplicate ACK, these are usually only
> sent as a signal of a lost segment [9] in regular TCP.  Therefore, an
> MPTCP implementation receiving a duplicate ACK which contains the
> MP_PRIO, ADD_ADDR or REMOVE_ADDR MPTCP option MUST NOT treat it as a
> signal of congestion.  Additionally, an MPTCP implementation SHOULD NOT
> send more than two duplicate ACKs in a row containing any of these three
> options, so as to ensure no middleboxes misinterpret this as a sign of
> congestion."

Yes it is clearer. 

Thanks for the background also. Based on your comments, I went over the ADD_ADDR section also and found this:
As discussed earlier, however, an MPTCP
   implementation MUST NOT treat duplicate ACKs with any MPTCP option,
   with the exception of the DSS option, as indications of congestion
   [9], and an MPTCP implementation SHOULD NOT send more than two
   duplicate ACKs in a row for signaling purposes.

So the statement you have or perhaps even the above same statement can be placed in the more prominent location of Section 3. That way, MP_PRIO is not lost in the discussion.
I am assuming that MP_FAIL and MP_FASTCLOSE do not count as they signal termination.  

> 
> 
> Olivier
> 
> -- 
> INL, ICTEAM, UCLouvain, Belgium, http://inl.info.ucl.ac.be