Re: [multipathtcp] A question related to MPTCP control overhead

[Olivier Bonaventure <[e-mail]>]

Hello,
>
> This is the first time I’m posting a question to the group!! I work for
> Huawei Technologies, India at Bangalore. For quite some time I’ve been
> reading material on MPTCP, and of late, I have been thinking of a
> potential issue related to MPTCP control overhead. I’m wondering whether
> the issue I articulated in the following is worth considering, and if
> so, I would love to hear opinion from the members of the group. I thank
> you all in advance!

Thanks for your email. Bascially, you would like to have a Multipath TCP 
connection that starts without DSS on this initial subflow and then 
enables the DSS when a second subflow has been established.

This would reduce the overhead when MPTCP connections are composed of a 
single subflow. However, in the presence of middleboxes, it looks very 
difficult to design a solution that would preserve connectivity under 
all circumstances.

Assume that MPTCP starts without any DSS option and only uses them after 
the establishment of the second subflow. If the bytestream has been 
affected by a middlebox (e.g. a NAT with an FTP ALG) that adds or remove 
bytes in the payload, when MPTCP will switch to using the DSS? they will 
be desynchronised with the actual sequence numbers that the receiver 
observers. There are other corner cases where DSS options are removed 
from packets with data but not from SYNs that could also result in 
various problems.

The idea looks nice, but there are unfortunately many devils hidden in 
the details


Olivier