Yggdrasil is an experimental compact routing scheme that is fully decentralised

https://yggdrasil-network.github.io/about.html

by doener • 1 year ago

194 46 1 year ago

yggdrasil-network.github.io

gnabgib• 1 year ago

Recently popular:

5 months ago (324 points, 107 comments) https://news.ycombinator.com/item?id=42155780

2022 (216 points, 78 comments) https://news.ycombinator.com/item?id=30156551

2019 as a Show (120 points, 15 comments) https://news.ycombinator.com/item?id=18863554

orbisvicis• 1 year ago

Is firewalling the yggdrasil adapter susceptible to IP spoofing?

gerdesj• 1 year ago

Some of us are old enough to remember a Linux distro called Yggdrasil. It was my first but certainly not my last.

Does this thing involve a MAC address?

beanjuiceII• 1 year ago

some of us even still have Yggdrasil installation media

smackeyacky• 1 year ago

I have a book and cd. I tried using the cd image with Qemu but I couldn’t work out how to get the CD interface going. It requires some ancient device emulation (I want to say the creative soundblaster CD interface) but I’m not sure the limited ones it knows about are in qemu.

makeworld• 1 year ago

AFAIK no MAC addresses are involved. It assigns you a new IP address, internal to the network, a bit like a VPN.

evbogue• 1 year ago

yah it's good for logging into computers or accessing servers behind NAT. It's kinda an antidote to this world that's still hooked on ipv4 like it's 1992 or whatever.

mindcrime• 1 year ago

Previously on HN (sadly no discussion though)

https://news.ycombinator.com/item?id=40828774

kachapopopow• 1 year ago

Why do I remember minecraft using yggdrasil at some point. Maybe just a name clash of a java library.

codetrotter• 1 year ago

Yggdrasil is from Norse mythology. So I’m not surprised that a bunch of different people name software after it.

https://en.wikipedia.org/wiki/Yggdrasil

Just like many might pick cool and important names from Greek mythology for example, like Poseidon.

https://en.wikipedia.org/wiki/Poseidon

privong• 1 year ago

Indeed; Yggdrasil is also the name of a set of models of the emission from stellar populations:

https://www.astro.uu.se/~ez/yggdrasil/ReadMe.txt

https://www.astro.uu.se/~ez/yggdrasil/yggdrasil.html

DigiEggz• 1 year ago

Brings to mind the system that rules over the Digital World :)

https://wikimon.net/yggdrasill

Centigonal3• 1 year ago

https://minecraft.fandom.com/wiki/Yggdrasil

awkwardpotato• 1 year ago

FYI, the Minecraft Fandom wiki is no longer maintained. The new wiki is https://minecraft.wiki/w/Yggdrasil

api• 1 year ago

The challenge with these schemes is always making them robust against intentional attacks.

It’s hard enough to do that for centrally regulated systems. There are whole giant companies like Cloudflare that mostly exist to do that.

With a fully decentralized mesh if someone finds a working attack there is no good way to coordinate a response.

Small hobby and research nets can get away with that, but if anything like this ever got popular or even close to mainstream it would be destroyed. This is especially true if there were any way to make money by ruining it.

nine_k• 1 year ago

Is there no way to blacklist nodes?

dexterdog• 1 year ago

Or throttle new node cration

Centigonal3• 1 year ago

Another related issue is that decentralized networks are often targeted by attackers more than centralized ones, since a common use case for these networks is sharing information that the central entity (or some power that can coerce the central entity) does not want shared, e.g. organizing in authoritarian regimes, illegal file sharing

fsflover• 1 year ago

Could you solve this by running something like I2P on top of the mesh network? I2P has implemented strong protections against typical attacks.

jazzyjackson3• 1 year ago

I don't know how ygg handles nodes but there's no reason a mesh network has to just accept random new peers, you could do something like mTLS where peers have to present certificates distributed out of band.

Animats• 1 year ago

The challenge with these schemes is always making them robust against intentional attacks.

Right. You join the network by generating a random IPv6 address and telling someone about it. So anyone can generate vast numbers of dummy node addresses, which everybody has to track. Now you need spam filtering.

sunshine-o3• 1 year ago

Yes but from my understanding one should not tell anybody about their IP as it is directly derived from a public key [0].

So in practice if you have doubts that your private key might have been compromised you must change your IP ASAP.

So I guess serious Yggdrasil users rely rather on a solid DNS system to manage identities and configure things. They actually need it more than the public internet !

This is the problem the blockchain & Ethereum people bumped into quickly: the 'key = identity' paradigm is elegant but in practice you need an abstraction on top to do anything serious.

I see they do have a DNS infrastructure in place [1] but most of the network services point to IP addresses, what really doesn't make sense to me.

- [0] https://yggdrasil-network.github.io/privacy.html#network-ide...

- [1] https://yggdrasil-network.github.io/services.html#dns

ogurechny• 1 year ago

There seems to be a misunderstanding. Yggdrasil is not a single network, it's a tool to make such networks. So if you have some computers with some ability to connect them, you can set them to peer in Yggdrasil, and instantly get an auto-configured irtual IPv6 network. (You can even slap some IPv4-to-IPv6 NAT over it, but we shouldn't mention crimes against humanity.)

It seems to me that the main goal of the project is automatic routing that promises that in any peer graph any peer can reach any other at least somehow, which is fine for many use cases. Only in case of moving big data or very different cost of links you might want to check the node hierarchy and set manual peering from X to Y and Y to Z to make traffic flow the way you want.

Public network is a volunteer-run test ground for real life usage, and a way to connect systems that can't connect directly because of NAT. Internal speed test that is 8 hops away on the other side of the world gives me 30 Mbps, closer one, in 5 hops, reaches 90 Mbps (and the local interface is 100 Mbps), so it has some throughput for casual users.

As far as I can understand, current latency-based routing selection is just the most simple way chosen to optimise the network across the whole world, and prevent random 200 ms hops back and forth to intermediate nodes. If traffic from nodes in one half of the tree not having direct connections to the other half goes through the central nodes, it seems that at the moment it's enough for owners of central nodes to peer with each other, and have enough bandwidth. In the future they might need to take the proportions of available channels into account.

bawolff• 1 year ago

Is there a paper somewhere explaining their system in detail?

drob518• 1 year ago

Yea, the website was conspicuously opaque about the protocol. Having lived in the networking and IP routing world for decades, that makes me suspicious that this thing is yet another non scalable system. But hey, I’m willing to be convinced.

easygenes2• 1 year ago

If this is interesting to you, there's also a new edge mesh protocol which would dovetail into it well called MeshCore. MeshCore is LoRa and MCU focused, works at L2. Ygg is L3 and could run at e.g. your LoRa SBC gateway to provide seamless global distribution of edge mesh chats and sensor relays.

https://github.com/ripplebiz/MeshCore https://youtu.be/fNWf0Mh2fJw https://chatgpt.com/share/681c281f-0a24-8011-8ec9-6d58ce3db0...

sunshine-o3• 1 year ago

I saw that with Reticulum, Yggdrasil is actually a "Backbone interface" [0] used to connect Reticulum instances. What makes a lot of sense.

- [0] https://reticulum.network/manual/interfaces.html#listeners

Ey7NFZ3P0nzAe• 1 year ago

Related discussion by reticulum devs: https://github.com/markqvist/Reticulum/discussions/162

eternityforest• 1 year ago

I'm stills waiting for the Meshtastic vs MeshCore vs Reticulum thing to shake out.... They all have pretty big disadvantages, most of which seem very fixable.

I think what they really need is cheap solar nodes that have external WiFi antennas, so they can mesh over point to point directional links.

LoRa meshing seems like it's eventually going to be a dead end in any kind of dense environments, WiFi just gets you so much more bandwidth, and the edge nodes only need ~2mA of power.

AFAIK only Reticulum and Meshtastic have software support for that ATM.

I have a proof of concept almost working using OpenDHT as a global routing backend for Meshtastic, but Meshtastic still has some security issues (No authentication on a lot of important stuff), and global routing is really only currently easy to do for unicast traffic.

But if they got rid of the channel hash thing, and instead they gave each channel a special receive address(like IPv6 multicast groups) with their own keypair, global routing would probably work fairly seamlessly.

I think ideally they'd change those node/channel IDs to rolling codes so you can't traffic analyze the global routing.

With MeshCore, it doesn't seem like there's documentation on what info the packets actually give you, so I'm not sure how that would work.

And then with Reticulum, there's no C++ implementation of LXML yet, no standalone devices, or anything like that, so I'm not sure if it can replace Meshtastic/MeshCore until that exists.

Meshtastic is also lacking some applications layer stuff, they have the same issue most IoT protocols have, tons of predefined packets for specific sensors, but no modbus-esque generic telemetry packets, no way to just say "Register 59 is 37.6, look up what that means in the register layout file with UUID Jfjeiuehbdhd"

I think Reticulum is closer to having that, with the "fields" thing.

I'm not sure how the global routing works on reticulum though, I get the impression every node needs to know about every other node with some kind of gossip protocol, but I could be completely wrong.

Reticulum also used Kivy for their mobile app, which doesn't feel as polished as Meshtastic, and it's licensed under some non-commercial creative commons thing that causes issues including it in distro repos and such, or so I've heard.

Meshtastic doesn't exist as a convenient Arduino library or Pip installable package or anything like that, it's not exactly very elegantly embeddable in existing apps, and even if it was, it doesn't have the generic abstractions to actually do custom stuff in a way the app understands.

It almost seems like we either need a heavily redesigned Meshtastic 3.0, or something completely new.

They're all amazing projects, but they're all missing something.

cyberax2• 1 year ago

Interesting. There's a fundamental vulnerability in mesh routing: you either have to control addressing, or a malicious actor can flood the network topology with bogus data.

Just imagine somebody generating 10000000000 addresses and flooding everybody with that information.

It looks like Yggdrasil doesn't address (ha) this vulnerability? It kinda side-steps it by requiring enrollment through an already trusted node?

mikeyg• 1 year ago

at least in the original implementation of this protocol (cjdns) there was already a tiny bit of proof of work happening in address generation. forgive me as this knowledge is over a decade old but it involved finding a curvecp private key that scalmults into a public key that (maybe whose hash?) begins with some specific number. this number is what's actually used as the ipv6 address. and that specific number is the routable ipv6 prefix.

the process would be run again and again during configuration generation until a key that fit this criteria was found. one could up the difficulty of this process considerably.. though not in a protocol backwards-compatible way.

you also needed to find a peer.

but yeah that's a gnarly hole.

cyberax2• 1 year ago

at least in the original implementation of this protocol (cjdns) there was already a tiny bit of proof of work happening in address generation.

It's just a 1 byte search, completely negligible from the performance standpoint. They look for a public key that has the SHA hash that starts with 0xFC, to indicate that it's not global IPv6 traffic.

I don't think it's possible to solve this, without either making a centralized addressing authority, or involving non-trivial amounts of real money via some blockchain.

getcrunk• 1 year ago

Mesh/decentralized networks seem poised to finally take off and become widespread. Unfortunately I don’t think it will be in the form of a cyber utopia free of control but a tool of power and surveillance, networking ai powered robots (swarm based weapons) and other smart things

pharos92• 1 year ago

Shortest path is fundamentally problematic in networks. The shortest path isn't always the best path.

The shortest path might be a 1Mbps link with high latency. The 'best' path might be several 100Mbps links with low latency chain together.

easygenes2• 1 year ago

Ygg doesn't operate strictly as a shortest path protocol. It prefers lowest latency links already and they're roadmapped to handle link cost / dynamic weighting / link congestion. They're targeting a blended weighted metric overlay which would certainly prefer the chain of quick/fat links.

jazzyjackson3• 1 year ago

I ask as someone hyped about decentralized protocol but only following ygg from a distance -- who are they building it for, is there a use case or commercial customer that wants this, or is it more of an idealistic/political endeavor ? I guess I'm wondering how mature the org behind it is, if they're setup for the long haul or taking steps forward when they can, on a hobbyist basis.

dhaavi• 1 year ago

I was intrigued by Yggdrasil, so much that I built a similar network, also using concepts of other networks in that realm.

It uses a logical world map for improved address space allocation and routing as well as source routing. Private non-routable addresses for better privacy are planned.

https://mycoria.org

It will see more development over the next years, as I will be using it in an upcoming project.

Happy to answer any questions you have here.

lostmsu• 1 year ago

What advantages does your project have over Yggdrasil?

randunel• 1 year ago

Is there a non-encrypted alternative to these things? There's lots of projects with services both paid and free, all having encrypted data transmission.

I'm looking for encrypted coordination with unencrypted data transmission for performance reasons in IoT devices, because the data itself is already encrypted (e.g. only https going through the tunnel).

theknarf• 1 year ago

How does this compare to Veilid (https://veilid.com/)?

nimoday• 1 year ago

which layer does it work on? layer 2 or 3?

Yggdrasil is an experimental compact routing scheme that is fully decentralised

Related Stories