Some news about Babel

[Juliusz Chroboczek]

Hi,

Discussions related to Babel are currently distributed across three
distrinct mailing-lists (babel-users at alioth, babel at ietf, bird-users), and
I'm a little concerned that those of you who are subscribed to just one of
them are missing out on the whole picture.  Since I'm on strike today, and
have some time before I go demonstrating, I've decided to to write up the
current activity.


1. IETF activity
================

We've managed to publish as RFCs almost all that's used in production.
The exceptions are:

  - relaxed MAC verification (draft-ietf-babel-mac-relaxed), which is
    stuck in limbo for reasons that I don't understand;

  - RTT-based routing (draft-ietf-babel-rtt-extension), which I keep
    wanting to rewrite, but keep getting sidetracked by things like
    student examinations and the French government implementing "reforms".


2. Babeld activity
==================

I've just released babeld-1.12.2, which is a pure bugfix release.  Babeld
is in a fairly stable state, which is good.

The master branch removes diversity routing (Babel-Z), which seemed like
a good idea at the time, but we never managed to show that it improves
performance over ETX.

There's work going on in a side branch to make redistribution more efficient,
I'm planning to release 1.13 when that is done.  No new user-visible
features are currently planned, sorry to everyone who requested.


3. BIRD activity
================

The implementation of Babel in BIRD has received a number of bug fixes and
new features, and it's slowly getting to feature parity with babeld.  This
is excellent news, since BIRD is an outstanding routing platform, and there
are a number of commercial routers that use BIRD as their control plane.

If you're running a production Babel network, I urge you to try out the
current master branch of BIRD (a version dated 2023-01-31 or later,
earlier versions have known bugs).  If BIRD works in your network, great,
you now have two independent implementations to choose from, and the bus
factor of your network has increased.  If BIRD is missing features that
you need, please contact the developers (most seem to subscribe to both
bird-users and babel-users), and I'm confident somebody will look at your
use case.

Just to be clear: the good state of the BIRD implementation does not mean
that I have any plans to abandon babeld.  It is important to have two
independent implementations of a protocol, it keeps the developers honest.


4. Major failures
=================

The big disappointment is that Microtik have still not implemented Babel
in their products (I was under the mistaken impression that they were
planning to do so when it became a standards-track RFC), and have instead
opted to implement yet another proprietary link-layer meshing protocol.

The other disappointment is FRR, whose implementation of Babel is based on
an old, buggy version of babeld.  The developers have accepted to fix some
of the bugs that I reported to them, then got bored and told me that
I could go fix the bugs myself.  I cannot currently recommend the use of
FRR for Babel.

The IETF Homenet working group has defined a protocol stack for home
networks based on Babel.  The Homenet protocols have been a technical
success, but a commercial failure, with no vendors that we know of
planning to implement the Homenet stack.  The Homenet working group is now
closed, and has been replaced by the SNAC working group, which has way
more modest ambitions.  SNAC does not mandate the use of a routing
protocol, which dramatically restricts what it can do (it is essentially
designed to connect a single Zigbee link to a home network).


Followups restricted to babel-users, but feel free to override my choice.

-- Juliusz