9fork ideas/high delay networking

[jrmu <jrmu@xxxxxxxxxx>]

Good morning mkf,

For international connections, high-delay networking is unavoidable.  If
there is a firewall and we're forced to use obfuscation tactics to smuggle
data, there is guaranteed to be high data loss and high latency. For the
next 5-10 years, I do not see any way to avoid this because of warfare
between US/EU and Russia/China/Iran.

Between borders, we may be forced to use any of four tactics:
1) DIY wireless (satellite and terrestrial) -- low throughput
2) sneakernet (think carrying microSD cards by car) -- high latency
3) steganography + vpns -- low bandwidth, high packet loss
4) vpns through neutral third parties -- unlikely

After deliberation, I think we may need our own 9fork. The direction we
need to go with 9fork will involve:

1) High delay networking
-- We can borrow ICANN/IETF/NASA's earlier work on interplanetary
networking with Bundle Protocol. I do not want to reinvent the wheel if
some prototype has already been tested and proven to work.

2) Steganography
-- It is insane to operate a vpn in Iran/China/Russia without
steganography. It will be dangerous even with steganography, but insane
without steganography.
-- This field seems relatively immature compared to cryptography -- it
will be a new opportunity for our network.

3) Federated networking
-- UNIX lacks federated internetworking protocols. 9fork is ideally suited
for this because 9fork makes an excellent pipe.  Here is a niche that
9fork can dominate in -- inter-firewall networking.  Systems programming
is where 9fork can beat unix

4) Store and forward protocols
-- Most of new networking done on Linux is focused on interactive
applications (matrix, webrtc, web apps, etc). These all break when round
trip time latency is measured in hours or days.

-- 
jrmu
IRCNow (https://ircnow.org)

Attachment: signature.asc
Description: PGP signature