F2F UDP mesh net prototype proof of concept

[Zenaan Harkness]

On Sun, Oct 20, 2019 at 08:40:09PM -0300, Punk - Stasi 2.0 wrote:
> On Mon, 21 Oct 2019 09:30:02 +1100
> Zenaan Harkness <zen at freedbms.net> wrote:
> > I could have called this RayzerNet, which has a better ring to it,
> 
> 	yep, but you don't need to call it anything. We can discuss the basic architecture without you using it as an excuse for your fascist propaganda.

I certainly don't need to call it "anything" - plain English words
are a bitch for search terms ... seriously.

Glad we agree on this one.


> 	So, what about defining only the core features of an hypothetical system? 
> 
> 	1) peer to peer - no 'directory authorities' 

ack


> 	2) are all nodes equal, bandwidth wise, or are there bigger nodes that provide some kind of 'convenient' (and less secure prolly) routing services?

all nodes equal, in the sense you use the word equal

every node is of course a unique snowflake - i.e. a unique set of
ISP, possibly >1 network connection, possibly >=1 dark fibre links
(off public network back links) such as N2N (neighbour to neighbour
ETH or WIFI down your street/ suburb), cpu, ram etc


> 	3) virtual-circuit-switched, or packet-switched? Is packet switiching the most expensive and the most secure option? Packet size? 

Now this is a good question.

We must consider the limits, and for sanity compare with existing
global Internet.

There are ~7 billion people on Earth.

The existing physical network (phys net), is clumpy - nations, in
particular island nations including Australia, have backbones running
around them joining cities, and branches out to regional areas, which
ultimately branch out to individual premises (a home, an office).

It's kind of a fractal of a star network, so sort of centralized.

At each aggregating point or "layer", is a router, routing within
that clump, and routing externally to other clumps, to other "star
networks".

With mobile phones (literally pocket supercomputers by the standards
of 3 decades ago), we have more dynamic possibilities, and demands
from users:

 - we -should- make use of physical peer to peer "ad hoc" wireless
   mesh networks, but this has yet to be solved in any production
   environment, except "sort of" by the occasional burning man etc
   set up - which have AIUI so far just been an "off grid" local star
   network replacing the traditional network

 - when known friends are near to each other, our phones should send
   text messages BYPASSING the national/ centralised govnet

Avoiding physical correlation of nodes is probably unwise, due to
such possible benefits. This implies inherent network "clumpiness".


So, back to routing - we can assume, with IPv6 at least, a flat
global node address space - at least, we can do this in a virtual
layer above govnet.

But do we want to?

We want to replace govnet, not assume govnet.

This means we want to in some sense create a practical physical alt
net, which is not govnet, and so our design MUST cater for this from
the get go, even though many links shall in the early days make use
of govnet.