2026-01-08 08:22:32

Recently, there has been an interesting technical discussion about Ethereum scalability—rather than solely pursuing ultra-low latency, increasing network bandwidth is actually a more reliable approach.

The reason is that with solutions like PeerDAS and zero-knowledge proofs, Ethereum's throughput could theoretically be increased by thousands of times. This level of horizontal scaling can ensure high efficiency without sacrificing the core principle of decentralization. However, latency is less flexible; it is directly limited by physical laws—if you want to accommodate global home nodes, ensure censorship resistance, and privacy, you can't push latency too low.

Therefore, the practical goal is to keep latency within the range of 2-4 seconds. For applications with higher real-time requirements—such as AI-driven interactions—Layer 2 solutions and localized "city chains" are relied upon. In other words, using different solutions for different scenarios is the most pragmatic approach.

The advantage of this thinking is to make full use of existing network resources rather than endlessly tinkering with the inherent constraint of baseline latency.

ETH-7,81%

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

NotSatoshi

· 01-11 08:12

I agree with the bandwidth priority argument; insisting on squeezing out latency is just stubbornly fighting against the laws of physics. --- PeerDAS is indeed impressive; it's much more reliable than speculative scaling solutions. --- 2-4 seconds is acceptable to me; after all, I’m not someone who plays games on the chain haha. --- The city chain idea is pretty good; it feels like a layered approach. --- Once again, the messianic theory of zero-knowledge proofs—will it work this time? --- I respect the principle that decentralization cannot be compromised; compared to those public chains that prioritize speed above all, they are more reliable. --- That’s right, different chains serve different purposes; insisting on a single underlying layer that supports all scenarios is truly foolish. --- Wait, what about those high-frequency applications now? --- Resource utilization optimization—finally, someone is giving it some serious thought.

View OriginalReply0

ForkItAll

· 01-10 14:52

Thousands of times throughput improvement sounds unbelievable, but PeerDAS is indeed a real deal. Its bandwidth optimization is much more practical than just reducing latency.

View OriginalReply0

GateUser-cff9c776

· 01-08 08:52

In plain terms, this is essentially admitting that Ethereum will never be able to do high-frequency trading. It sounds pragmatic, but it's actually just another way of saying "we give up." However, on the other hand, PeerDAS combined with zero-knowledge proofs is indeed the most effective solution at the moment. The throughput increase by thousands of times should, according to traditional financial valuation logic, be reflected in the decrease of gas fees—if this can truly be achieved, then those L1s claiming to be Ethereum killers might find themselves unable to sit still. A delay of 2-4 seconds is actually quite realistic; after all, the immutable laws of physics are deeply ingrained in my mind. Instead of fantasizing about instant confirmation, it's better to make good use of bandwidth to build the ecosystem.

View OriginalReply0

ponzi_poet

· 01-08 08:40

Bandwidth routing is indeed reliable, much more practical than those schemes claiming ultra-low latency. --- It's the same old physical laws; we've long since realized this. --- I'm optimistic about PeerDAS; a thousandfold throughput isn't a dream. --- Layer2 can't solve certain problems just by reducing latency; layering is the real way to go. --- The city chain concept is good, but I'm worried it will just become another gimmick to scam the chives. --- Be more realistic, don't always think one scheme can conquer everything. --- Decentralization and low latency are truly incompatible; this guy finally explained it clearly.

View OriginalReply0

FrogInTheWell

· 01-08 08:29

I have to say, this approach is indeed clear-headed. I never believed in those schemes that claimed "millisecond-level latency," as they violate physical principles. Separating bandwidth and latency, and using Layer 2 for traffic diversion, is a more practical approach. It’s much more reliable than artificially reducing the fundamental latency. If PeerDAS can really be implemented, the potential for thousands of times increase in throughput... is a bit exciting. Wait, is 2-4 seconds really unavoidable? It still feels a bit painful. It's all about trade-offs—decentralization and performance, the classic adversaries.

View OriginalReply0