The propagation method of public chain networks is actually a performance bottleneck that is easy to overlook.
Many people's imagination of public chains still stays at this stage: receiving a message and forwarding it to all neighboring nodes — this is a typical flood/gossip mode. It sounds simple and crude, but there are many issues. Messages are repeatedly forwarded, redundancy coverage is severe, and bandwidth leaks out like a floodgate opening. The more nodes there are, the more obvious the collisions and congestion become. The final result is: slower confirmations, unpredictable delays, and network-wide congestion causing everyone to freeze.
Some public chains choose a different approach. Instead of "forwarding to everyone," they select forwarding targets based on XOR distance in a Kademlia DHT. Each node only sends messages to a set of "progressively farther" selected nodes, allowing information to cascade across the entire network like a waterfall.
You can think of it this way: it's similar to a modern logistics system. Packages are not sent from point A to all other points directly, but routed through the most suitable transfer stations based on routing and hierarchy, spreading layer by layer. This approach can cover the entire network while precisely controlling costs and efficiency.
The benefits of this "structured distribution" are quite straightforward. First, redundant forwarding is greatly reduced, significantly lowering bandwidth pressure. Second, the probability of message collisions decreases, making propagation delays more controllable and stable. For blockchain scenarios that require serving regulated financial environments or are highly sensitive to delays, optimizing from the network layer is crucial.
The "second-level experience" that ordinary users feel is never just about how fancy the consensus algorithm is. The real underlying skill starts from the network layer — delivering each block, each vote, and each transaction faster and cleaner to where it should go. This broadcasting mechanism is truly like a "route-based logistics," rather than a "shouting in a square."
View Original
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
12 Likes
Reward
12
5
Repost
Share
Comment
0/400
GasFeeCrier
· 5h ago
It seems to be the old issue of P2P network design, gossip mode is indeed prone to congestion.
---
DHT routing has been used by some people for a long time, so it's not really new.
---
That's quite right, if the underlying network isn't optimized, even the fastest consensus is useless.
---
So essentially, it's about optimizing network topology, just changing the forwarding strategy.
---
How are the chains that actually use DHT doing now? Are their performance really that much better?
---
The square with shouting haha, perfectly described, it's about the current state of most chains.
---
The problem is most node operators don't care about this anyway, as it costs nothing.
---
Wait, does this have anything to do with the inherent design limitations of Kademlia?
---
Sounds like it's educating projects still using primitive gossip.
---
What are the costs of structured distribution, has that been discussed?
View OriginalReply0
FOMOSapien
· 5h ago
The analogy of the bustling square is spot on. Right now, a bunch of public chains are just a chaotic mess.
View OriginalReply0
MetaverseHermit
· 5h ago
Haha, the square is awesome. I used to think TPS was everything, but I didn't realize how crucial the network layer is.
View OriginalReply0
IronHeadMiner
· 5h ago
The analogy of the square is spot on; it truly reflects the current state of most public chains.
We should have paid more attention to the network layer long ago, otherwise even the strongest consensus is useless.
The last mile is often the easiest to fail; the true test lies in the details.
View OriginalReply0
HashRatePhilosopher
· 5h ago
Oh, so that's the real reason those old chains keep lagging. No wonder transfers move like turtles.
---
Structured forwarding is indeed impressive, but how many chains are actually using it?
---
Basically, it's about optimizing routing. Why are so many projects still using the outdated gossip protocol?
---
The logistics analogy is spot on. Finally, someone explained the network layer so clearly.
---
Kademlia DHT has been around for a long time. Why is it only now being valued?
---
After some thought, this explains why those high-speed chains are really fast—it's not just due to the consensus algorithm.
---
The problem is that implementing this requires changing the underlying code, and most chains are too lazy to do that.
---
The real key to a second-level experience is in the details. Small optimizations, accumulated over time, lead to a qualitative leap.
The propagation method of public chain networks is actually a performance bottleneck that is easy to overlook.
Many people's imagination of public chains still stays at this stage: receiving a message and forwarding it to all neighboring nodes — this is a typical flood/gossip mode. It sounds simple and crude, but there are many issues. Messages are repeatedly forwarded, redundancy coverage is severe, and bandwidth leaks out like a floodgate opening. The more nodes there are, the more obvious the collisions and congestion become. The final result is: slower confirmations, unpredictable delays, and network-wide congestion causing everyone to freeze.
Some public chains choose a different approach. Instead of "forwarding to everyone," they select forwarding targets based on XOR distance in a Kademlia DHT. Each node only sends messages to a set of "progressively farther" selected nodes, allowing information to cascade across the entire network like a waterfall.
You can think of it this way: it's similar to a modern logistics system. Packages are not sent from point A to all other points directly, but routed through the most suitable transfer stations based on routing and hierarchy, spreading layer by layer. This approach can cover the entire network while precisely controlling costs and efficiency.
The benefits of this "structured distribution" are quite straightforward. First, redundant forwarding is greatly reduced, significantly lowering bandwidth pressure. Second, the probability of message collisions decreases, making propagation delays more controllable and stable. For blockchain scenarios that require serving regulated financial environments or are highly sensitive to delays, optimizing from the network layer is crucial.
The "second-level experience" that ordinary users feel is never just about how fancy the consensus algorithm is. The real underlying skill starts from the network layer — delivering each block, each vote, and each transaction faster and cleaner to where it should go. This broadcasting mechanism is truly like a "route-based logistics," rather than a "shouting in a square."