Hacker News Reader

Comment by GistNoesis | Hacker News Reader

Note here that "faster" here really means more speed and not an increase in the volume of data transferred : The light go through the air hollow-core so can go at near "c" (the speed of light in vacuum) speed, instead of being constrained to "speed of light in glass which is only "2/3 c". This allows reduce latency for long distance communication.

https://spie.org/news/photonics-focus/julyaug-2022/speeding-...

View on HN

It's true, but for most cases, the volume of a fiber is not the problem anyway. Latency is a problem most of us somehow bump into every day, while most fibers in the ground are nowhere near what you can push out of DWDM (e.g., off-the-shelf equipment will easily allow you to run 20x100Gbit over a single fibre, but many of them only carry a single 10Gbit or even 1Gbit link).

Trans-continental is different, because you'll need amplifiers. Many, many amplifiers in a row. And those generally work well only in a fairly limited band. But unless you're doing submarine, bandwidth is almost never the problem.

To make things worse, a lot of existing medium-haul fiber links are actually twice as long as you'd expect, due to the desire to cancel out dispersion; you first run the fiber e.g. 10km from place to place, and then run it through a large 10km spool (of a slightly different type of fiber) in the datacenter to cancel out the dispersion. This is slowly going away, but only slowly.

>To make things worse, a lot of existing medium-haul fiber links are actually twice as long as you'd expect, due to the desire to cancel out dispersion; you first run the fiber e.g. 10km from place to place, and then run it through a large 10km spool (of a slightly different type of fiber) in the datacenter to cancel out the dispersion. This is slowly going away, but only slowly.

Only slowly? I haven't worked with very long haul wdm systems or sonet/sdh but I've never seen this. Maybe you mean much longer distances than 10km as we've had 10G-LR for ages that don't need this.

I don't know exactly how long gives you too much dispersion (obviously depends on the fiber), but if it's short enough you just don't need to care, indeed.

I don't work with this myself, but my understanding is that you only start ripping this out when you are positive every wavelength from every customer actually has coherent detection, and that could take a while. :-) Obviously this will differ from site to site, too; I would assume new ones don't care unless there's a lot of legacy involved.

Alien wdm isn't common in long haul systems though, if so the provider typically provides the transponder. Usually customers use regular grey (1310nm 10G-LR) to the provider and the provider turns that into something compatible with their dwdm system.

Huh, interesting, I didn't know this was a thing at all. But you still need to make sure you don't have a stupid customer sending 1000baseLX with direct detection in the other side for this to work. :-)

Newer coherent optical transceivers don't need physical dispersion compensation as they just do it in digital signal processing instead. That is the magic of ever increasing rates of computation at lower power expenditure.

Thus going away (see my other comment in the same subthread).

Sounds like it was something like CML for >200km runs. =3

does using hollow core means you can do away with dispersion compensation?

AFAIK yes, but if that's your goal, a far easier solution is to just use transmission standards that don't care about dispersion (coherent detection). E.g., all 100gig transmission already does not care about it.

This is actually much more important than the volume of data transferred. Having 33% lower latencies across the globe would be huge.

On the other hand, it is only 33% - and that is an upper bound.

Getting data to literally the other side of the globe currently takes about 100 milliseconds. How many truly novel applications open up by that latency dropping to 66ms?

For short-distance stuff the latency is already low enough to be practically realtime. For long-distance stuff we're already fast enough for human-level applications (like video chat), but it's not dropping enough for computer-level applications (like synchronous database replication).

I'm sure some HFT traders are going to make an absolute fortune, but I doubt it'll have a huge impact for most other people.

I made my master thesis on real-time, with a chapter where I experimented with different levels of jitter and latency. Jitter is the consistency of the latency, is it like a locked 66ms or sometimes does it go to 200ms. Jitter is more impactful than latency for a wide range of applications, from gaming to music and video call. Having a lower latency allows for lower jitter, or less jitter while keeping the same latency. Today’s discovery is huge imo.

Doesn't jitter come from the switches and routers along the path? I have a hard time seeing a fibre having significant jitter.

Also, there is a very narrow threshold of latency timings in which "real time" communication goes from looking real time to actually feeling real time. That narrow window is why people end up interrupting each other or feeling like they can't get a word in edge wise on video calls all the time.

> I'm sure some HFT traders are going to make an absolute fortune, but I doubt it'll have a huge impact for most other people.

They’ve been using hollow core fiber (and funding research into it) for nearly a decade. I know it goes back further than the 2017 spinoff mentioned in the article, but https://optics.org/news/11/9/52 talks about it a bit.

I generally agree with you, but! Video or audio calls between EU and the US still have a much higher chance of speaking up at the same time and it’s due to lag. If the latency is decreased by 33% it might be a game changer.

Mind-boggling logic, for example any existing roundtrip-heavy application (such as CIFS) would gain visibly and immensely because that latency is multiplicative

online music playing is HEAVILY latency sensitive. (for instance an online jazz session)

then you have online video games. increasing the area where you can get good connections increase quadratically (or more, if we hit step function = big city get in range) the viability of niche multiplayer video games and it is thus a boost to creativity.

there are probably many more niches... (need to think of reachable area, quadratic, instead of 1-to-1 link linear)

> and that is an upper bound

I've often wondered if for HFT or similar it might be worth pointing a particle accelerator at the floor and going for direct-line transit times. I'm fairly sure that this is theoretically possible, but no idea if the engineering challenge is beyond reach for use as a communication link.

If your signal is "transparent" enough go through so much rock and iron without being absorbed (like neutrinos), you'll have a hard time capturing it on the receiver side.

Well, OPERA was 700ish km, but had Cern at one end. If one has this as the sole goal and wanted to do it real-time over 12,000km is it "engineering-possible" vs "theoretically-possible" ? My guess is that it depends how much money stands to be made ;)

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

Basically just aim you accelerator at any neutron detector.

Problem is you'd drop more packets than IP over pigeons.

I think you're confusing neutrons and neutrinos. Firing neutrons at the floor will definitely give you a very radioactive floor.

Does carrier pigeons have that high packet loss?

HFT traders probably won't make any extra money unless they deploy this first to their dedicated international links.

Almost all of them deploy their strategies within exchange colo's already

There's a lot of need for communication still. In US, futures trade mostly in Chicago, but equities in New York, for example. In Europe things trade all over the place.

What this would do is increase the radius of where you can do some latency constrained thing. If your latency budget is 20ms, you could now do that over a bigger areas.

HFT uses microwaves for anything over distance. Unless this beats microwave latency, this doesn’t do anything for them fwiw

Microwave is only feasible over medium distances - can't do it over the ocean, as it requires LoS. Also IIRC, microwave bandwidth is considerably lower than fibre, and sometimes it matters.

Microwave is also affected by weather. They sometimes say that markets are slightly less efficient on rainy days. It’s a bit of a joke, but basically packet loss goes way up and they rely more on fiber links when microwave links are being shitty.

Traders have at least experimented with shortwave too: https://spectrum.ieee.org/wall-street-tries-shortwave-radio-...

You still need to traverse physical segments in the wireless path: think receiving dish to the next transmitting dish, the end of the path to get from the trading systems onto the roof and into to the first dish, etc. Every nanosecond counts.

Great for gamers

don't know about truly novel, but CS:GO players certainly would benefit.

Thanks for your precision. Off-topic: It's true that "faster Internet" means "bigger Internet" in common language, just like in photography "faster lens" means "lens with more light gathering capability". I wonder if there's other field where "faster" is misused.

Higher bandwidth means less time to transmit a given packet. Just like more light gathering in a lens allows faster shutter speeds.

If you had to move 100 tons of packages, which is going to be faster - a Lamborghini going 200 mph, or a tractor/trailer going 50mph?

If you’re trying to set a speed record and don’t care about bringing anything along, which is faster?

Neither meaning is necessarily wrong.

Even if directly it isn't technically right to say it's faster, but with it's applications built upon such technology it manifests in these tasks being done faster. Makes sense.

And the time for the lambo and the tractor will depend on the round trips each will have to do, so it depends on the medium.

So submarine cables were limited to 0.67c and now aren’t? Can this really work?

For submarine cables there are two things here. The first is lower attenuation which allows for fewer amplifiers along the route making it overall cheaper. The second is lower latency. There have been cases where high frequency trading people went wireless to get lower latency because of the higher propagation speed of EM-waves in air. For really long distances you can go theoretically use satellite links to get lower latency than a submarine cable even if the total distance increases.

Someday, someone is finally going to work out how to do comms with neutrinos (which can pass directly through the Earth and come out the other side) and make so much money...

They are not, for the thing that would make neutrinos useful for communications is also the thing that makes them useless for communications. In order to use them for comms you'd need to produce such a huge number of neutrinos, and/or in a very colimated beam, that one shudders to think of how one might produce them!

I think you could moderate a large nuclear fission reactor (on the edge of criticality) to produce detectable differences in neutrino rate on the other side of the world. KamLAND and a few other experiments detected multi-GW reactor anti neutrinos at 1000km[0] and so presumably tens of GW (easily [if perhaps not safely?] achievable briefly in current reactors) should be detectable over 8000km.

[0] https://www.sciencedirect.com/science/article/pii/S173857331...

I suspect it may be possible using neutrino-antineutrino annihilation if there's a good way to produce streams of both types of particles, but... physics isn't my field.

> So submarine cables were limited to 0.67c and now aren’t?

I think it’s more like in the future they might not be. It’s anyone’s guess how mass production and deployment of this might look.

Reply on news.ycombinator.com