Here's a summary of the themes from the Hacker News discussion, with direct quotes:
Direct Copper vs. Fiber Latency at Short Distances
A central theme is the surprising finding that direct copper cables are faster (lower latency) than fiber optics for very short distances (under 5 meters). This contradicts a common assumption that fiber is inherently faster in all networking scenarios.
The Role of the PHY Layer
The Physical Layer (PHY) circuitry, which handles signal regeneration and conversion, is identified as a significant factor contributing to the latency difference in short-reach connections. When the distance is minimal, the time spent by the PHY processing the signal outweighs the speed advantage of the transmission medium itself.
- laurencerowe questioned, "Surely resignaling should be the fixed cost they calculate at about 1ns? Why does it also incur a 0.4ns/m cost?"
- vlovich123 elaborated, "Faster only because the distances involved are short enough that the PHY layer adds significant overhead."
Speed of Signal Propagation (Velocity Factor)
The discussion delves into the physics of signal transmission, explaining why copper can be faster than fiber at short distances. This boils down to the speed of electrical versus optical signals within their respective media, often referred to as the "velocity factor." While light in a vacuum is the ultimate speed limit, neither electrical signals in copper nor light in fiber travel at that speed.
- cenamus suggested, "Light speed is ~3ns per metre, so maybe the lowered speed through the fibre? Speed of electricity in wire should be pretty close to c (at least the front)"
- b3orn clarified, "It's c, but not the same c as in air or vacuum. The same applies in optic fibers. They're both around two thirds of the speed of light in vacuum."
- Sesse__ provided more precision: "c is the speed of light in vacuum. EM signals move at about 0,66c in fiber, and about 0,98c in copper."
- myself248 added, "Velocity factor in most cables is between 0.6 and 0.8 of what it is in a vacuum. Depends on the dielectric material and cable construction."
- laurencerowe corrected a prior misconception: "I misremembered the speed of electrical signal propagation from high school physics. It's around 2/3rds the speed of light in a vacuum not 1/3rd. The speed of light in an optical fibre is also around 2/3rds the speed in a vacuum."
- zokier identified the source of the per-meter difference: "The PHY contributes only 1ns difference, but the results also show 400ps/m advantage for copper which I can only assume to come from difference in EM propagation speed in the medium."
- myself248 further explained the graph interpretation: "No. Look at the graph -- the offset when extrapolated back to zero length is the PHY's contribution. The differing slope of the lines is due to velocity factor in the cable. The speed of light in vacuum is much faster than in other media. And the lines diverge the longer you make them."
- nimos summarized, "This isn't really surprising. Fiber isn't better because of signal propagation speed, it's all about signal integrity."
Real-World Implications and Future Trends
The discussion touches upon whether these latency differences have practical consequences and how technology is evolving. Error rates, crosstalk, and the obsolescence of PHYs are considered.
- MadVikingGod raised concerns about practical issues: "What I would actually like to see is how this performs in a more real world situation. Like does this increase line error rates, causing the transport or application to have to resend at a higher rate, which would erase all savings by having lower latency. Also if they are really signaling these in the multi GHz are these passive cables acting like antenna, and having a cabinet full of them just killing itself on crosstalk?"
- palomides offered a counterpoint regarding error rates: "high speed links all have forward error correction now (even PCIe); nothing in my small rack full of 40Gbe devices connected with DACs has any link level errors reported"
- tcdent predicted a shift: "PHYs are going away and fiber is going straight to the chip now, so while the article is correct, in the near future this will not be the case."
- hilift noted the convergence where 10 Gbps copper and fiber are similar, but fiber scales to higher speeds: "Storage over copper used to be sub optimal but not necessarily due to the cable. UDP QUIC is much closer to wire speed. so 10 GB copper and 10 GB fiber are probably the same, but 40+ GB fiber is quite common now."
Nuance in the Term "Faster" (Latency vs. Bandwidth)
There was a recognition that "faster" in the context of the article was referring to latency, not necessarily bandwidth, which led to some initial surprise.
- jerf commented, ""Has lower latency than" fiber. Which is not so shocking. And, yes, technically a valid use of the word "faster" but I think I'm far from the only one who assumed they were going to make a bandwidth claim rather than a latency claim."
- kragen shared a similar reaction: "I assumed they were going to make a bandwidth claim and was prepared to reject it as nonsense."
Specific Use Cases for Low Latency
The discussion identifies specific industries and applications where even nanosecond-level latency improvements are critical.
- throw0101d highlighted the relevance for financial markets: "This specifically mentions the 7130 model, which is a specialized bit of kit, and which Arista advertises for (amongst other things): > Arista's 7130 applications simplify and transform network infrastructure, and are targeted for use cases including ultra-low latency exchange trading, accurate and lossless network visibility, and providing vendor or broker based shared services. They enable a complete lifecycle of packet replication, multiplexing, filtering, timestamping, aggregation and capture. [...] Latency matters (and may even be regulated) in some of these use cases."
- zokier asked for further examples, "What are applications where 5ns latency improvement is significant?"
- thanhhaimai responded directly, "High Frequency Trading is one."
- Loughla queried, "Anything else? Because that's the only one I can think of."
Arista's Niche and Founding
The discussion briefly touches on Arista Networks' historical focus on low-latency solutions, particularly within the financial sector, and mentions its prominent co-founder.