this post was submitted on 19 Jul 2023
18 points (100.0% liked)

networking

2845 readers
2 users here now

Community for discussing enterprise networks and the ensuing chaos that comes after inheriting or building one.

founded 2 years ago
MODERATORS
 

Hiya Folks

Making the rounds again on this project as it is getting closer to being feature-complete (ish) and I've started this website for extended infoz/documentation. Main repository can be found here: https://github.com/webmeshproj/node.

The project aims to be yet another simple WireGuard Meshing/VPN solution. Most similar to TailScale/HeadScale, but with a controller-less architecture governed by Raft consensus.

I'm excited to hear any feedback. Contributions are welcome as well :). Anything from architecture discussion, to issues, to code, to docs is appreciated.

you are viewing a single comment's thread
view the rest of the comments
[–] [email protected] 1 points 1 year ago* (last edited 1 year ago) (2 children)

I would be very interested in this! What do you mean by raft consensus? Ah, never mind. I used Google. Still I would be very interested because I cannot see myself using a corporate provider like Tailscale and I like the idea of a controller-less setup.

[–] tinyzimmer 2 points 1 year ago

I am by no means an expert but the TLDR is Raft is a protocol that allows distributed systems to maintain a central state. The GitHub page on it is pretty good - https://raft.github.io/.

What it means for this project is that every single node keeps the database containing the entire network state (rules, addresses, routes, etc.) in-memory. At any single point in time, any of the "voting" nodes can become the "leader". The leader is responsible for authorizing nodes to join, mutating state, etc. If that leader goes away - another node will pick up the slack.

[–] tinyzimmer 1 points 1 year ago* (last edited 1 year ago)

Hehe I'll respond to the edit.

I actually have a lot of respect for what TailScale is doing. 99% of their shit is open source and they don't get in the way of the downstream Headscale project that lets you run your own controllers. That being said, I think it gets pricey at scale and tries to do too much for the user. Extending it isn't super easy at the moment either, but they are working on ways of embedding their agents.

I wanted to take the idea and put it on the same level of distributed internet projects like Reticulum. I think this has potential to be the networking base for a concept similar to "dApps" but removing the financial incentives that come with using blockchain.

That all being said - I'm totally considering making a managed offering of this - and am actively looking for people who'd be interested to go on that journey with me. But I'd try extremely hard to never be labeled "corporate" :P.