MPTCP (Multipath TCP) support/testing about go-any-proxy

[ryanchapman]

Sorry for the delay; GitHub didn't email me when you opened this issue. If the kernel on N just needs to open a TCP socket to S for each path, then you could set up a vyatta machine running any_proxy that serves as the default gateway for N and it should work. But do you need any_proxy? It's intended for giving machines access to the internet that don't have a direct route. This is common in large corporations where they don't allow machines to make connections directly to the internet; instead, they force machines to connect through a proxy server like squid.

[dumblob]

But do you need any_proxy?

In principle not. The question starts to get interesting though once you're moving all the time in the whole continent with the notebook and can't have any other box with vyatta running on it (well, the virtual appliance, which could run directly on the notebook, is also an extreme overkill).

Proxies in general are built around a highly efficient bidirectional reliable communication between two network nodes. So I just wanted to know, whether any_proxy would support MPTCP. This usually requires some work as MPTCP extends the standard unix sockets - see ftp://ftp.fu-berlin.de/doc/rfc/authors/rfc6897.xml#rfc.section.3.2.2 and ftp://ftp.fu-berlin.de/doc/rfc/authors/rfc6897.xml#sec_query_addr (IPs may be added and removed dynamically very often) and ftp://ftp.fu-berlin.de/doc/rfc/authors/rfc6897.xml#sec_basic_sockets (new API supporting the dynamics of addition and removal of subflows).

[ryanchapman]

TODO for anyone wanting to send in a PR: look into feasibility of MPTCP within any-proxy