[ZS] Re: looking into cjdns

[Bryce Lynch]

On Wed, Oct 10, 2012 at 10:22 PM, Spencer Campbell
<lacertilian at gmail.com> wrote:

> My impression is that Byzantium's ad-hoc mesh network would look a whole lot
> like the networks we're used to, though. If it grew big enough. Am I

At OSI layers 3 and above, it is.  TCP/IP is TCP/IP.  At OSI layers 1
and 2, it uses the ad-hoc mode specified by the 802.11 spec rather
than infrastructure/managed mode (which is basically a lots of
clients/one access point star topology).

> mistaken? Maybe Byzantium doesn't scale the way I've assumed it does, and

We're not sure how big it scales, to be honest.

We're not concerned about how the IP routing algorithm is going to
scale - those are well known and understood, and already scale to a
couple of hundred nodes if not more.

One of the biggest limitations of wi-fi in ad-hoc mode is range -
consumer wi-fi is good to a couple of hundred meters at best.  The
other is that there are vendors out there who are actively hostile to
ad-hoc mode - Google among them.  At the International Summit for
Community Wireless Networks in Barcelona, we wound up doing a
presentation on it and found out from a lot of other projects what a
serious problem it's been (search terms: "android bug 82").

Now, for what it's worth it doesn't actually take a lot of nodes to
cover a lot of space.  When we were field testing at CarolinaCon we
covered an entire parking lot with a single unmodified node.

> you'd end up with a sort of "clumpy" situation, where you have less of a
> global Internet and more of a smattering of smaller imperfectly-connected

That is correct, and that's pretty much what we expected.  That is why
our problem space involves emergency first response and infrastructure
failures.  We need to finish reworking things before we can start
planning Battlemesh DC to do a real field test.

> internets. I guess that's implied by the emphasis on wireless connections --
> hard to get WiFi across continents. It seems clear that cjdns is intended to
> scale to arbitrary sizes. So, the infrastructure goes down and you rely on

The thing about cjdns is that it relies upon existing infrastructure,
which might be hostile.  For example, we know that Bluecoat
net.censorship equipment has been blocking the cjdns protocol since
April of 2011, and some very interesting intel came out of it.
Summarizing the presentation at HOPE 9 that a few of us from Telecomix
gave:

Bluecoat in California has a processor cluster that is dedicated to
analyzing samples of network traffic, generating rules to a) detect
and b) block new protocols, and pushes them to every Bluecoat device
on the planet after automated unit testing (this is one of the reasons
we knew they were lying when they said that they didn't know their
boxen were deployed in the Middle East, we watched them push updates
to the machines).  Total turnaround time from deployment of cjdns to
blocking of cjdns: About five minutes.  It's not usable from any of
the work sites I go to that use Bluecoat gear, either and hasn't been
since that time.

I have no idea if cjdns is blocked by any other censorship technologies.

So, cjdns is useful until the infrastructure decides to make it unusable.

> Byzantium until the new infrastructure (based, perhaps, on cjdns) comes
> online? Is that the idea?

cjdns isn't infrastructure - it's not a switch, a router, a network
card, or anything like that.  Let's get terminology settled.

Byzantium is designed to turn commodity computers into infrastructure;
IP routing is performed, packets are forwarded (the Linux kernel is
surprisingly good at that), and interfaces are used for the relay of
traffic.  I don't see why a cjdns net couldn't be bootstrapped on top
of a mesh running Byzantium, but I have some concerns about the two
routing protocols not being compatible with one another.

> Seems like it'd be more efficient to just give the emergency system the
> ability to "grow up" into a new status quo. Then I wouldn't have to install

That's sort of what we have in mind. :)

> a new distro when the post-post-apocalypse rolls around. You have to be
> practically there already, given that Byzantium interfaces with the
> contemporary Internet without complaints, yeah?

Pretty much.  That's also one of the reasons we're building
application frameworks into Byzantium, so it's possible to run
arbitrary servers on top of Byzantium in addition to just pushing
packets around.

> (Thanks for humoring me. I hope my profession of ignorance is less painful
> for you than it is for me.)

Don't worry about it.  It's not nearly as painful as this was:

http://www.washingtoncitypaper.com/blogs/housingcomplex/2012/01/11/meet-the-new-ish-boss-chief-technology-officer-rob-mancini/

-- 
The Doctor [412/724/301/703] [ZS]
https://drwho.virtadpt.net/
"I am everywhere."

-- 
-- 
Zero State mailing list:
http://groups.google.com/group/DoctrineZero



----- End forwarded message -----
-- 
Eugen* Leitl <a href="http://leitl.org">leitl</a> http://leitl.org
______________________________________________________________
ICBM: 48.07100, 11.36820 http://www.ativel.com http://postbiota.org
8B29F6BE: 099D 78BA 2FD3 B014 B08A  7779 75B0 2443 8B29 F6BE