Greg's Cable Map | Hacker News Reader

Greg's Cable Map

cablemap.info

83 pointsajr041 comments

The same map in 1901

https://upload.wikimedia.org/wikipedia/commons/a/a5/1901_Eas...

Actually, not much changed. There's probably less active cables now.

There's a lot more in the Pacific nowadays than on that map. And I don't think less cables are active. Our need for bandwdith nowadays compared to 1900 is much higher, a lot of those cables have simply been upgraded from carrying telegraph signals to snappy fibre optics.

I'm surprised at how many cables crossed the Atlantic in 1901. Anyone know how they power the amplifiers under the ocean? (Then and now?).

Edit: It appears that there's a separate wire carrying electricity. I'm no electrician but I was under the impression that power loses it's Voltage (or Amperage?) over distance.

See https://en.wikipedia.org/wiki/Submarine_communications_cable... for a brief overview.

Early submarine cables were literally just wires. To overcome the electrical resistance of a long cable, they simply applied a large voltage on one end, and used sensitive instruments to pick up the tiny signal on the other end.

Interestingly, the early engineers didn't fully understand how the electromagnetic properties of a cable (capacitance and inductance) affect signal propagation over long distances. So they ended up seeing unexpected amounts of distortion that forced them to transmit much more slowly than desired. The theoretical breakthrough that enabled reliably sending signals from one side of an ocean to another was thanks to Oliver Heaviside, who was pretty much the archetype of a misunderstood crackpot/genius.

But yeah, later cables included in-line repeaters, once the technology was sufficiently well developed. Here's a description of some early vacuum-tube amplifiers that were designed for extreme reliability: http://www.ase-museoedelpro.org/Museo_Edelpro/Catalogo/tubes...

Power does diminish with distance due to I2R losses. For cables using AC, there are also reactive losses. To compensate, the cable companies boost the voltage at both ends (30-500 kV) of the cable such that the repeater at the middle of the span has enough juice to operate properly. And, yes, separate cables, or wire bundles, carry the power and ground.

Chapter 11 of The Idea Factory by Jon Gertner (a book about Bell Labs) has a fun story about the first transatlantic telephone cable, completed in 1956. According to him, telegraph cables were attempted as early as the 1850s, but the first successful one was in 1866, between Canada and Ireland.

What is the island(?) in the middle of the Atlantic where 5 cables meet? I can't quite make it out from the image.

The Azores

The Azores?

Google Cache for those in the "connecting" state. http://webcache.googleusercontent.com/search?q=cache:_BoKori...

I was in Hawaii this weekend and explaining this to my girlfriend who is not at all a "techy". I explained to her, how odd it is, that the internet works here because there's literally a giant cord from the island back to California laying on the ground in the ocean.

Seems rather primitive when you stop and think about it.

what are the options? Satellite transmission certainly is one but it is bandwidth limited and it imposes delays due to the distance involved. Microwave radio won't work because it's line-of-sight. HF or VHF radio has atmospheric effects and noise issues and is even more bandwidth constrained. So, really, cable is the best option giving the lowest price per bit. Fiber, these days, is approaching 100 Gbps with the ability to multiplex multiple bit streams onto different wavelengths to allow transmission of insane amounts of data. Hardly primitive when you do think about it.

There hasn't been any other option. However, I think we could launch a low orbit satellite constellation, now. It will be exciting to see how well it will actually work.

You mean like Iridium?

> Satellites are in low Earth orbit at a height of approximately 485 mi (781 km) and inclination of 86.4°

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

You can get Irudium or Globalstar today in 9600 bps of glory with rain and cloud fade issues.

...and the less is spoken of the time I found myself >1000km from the nearest wired network point and needed to download an 8MB file over Iridium, the better.

Let's just say it makes you appreciate download managers a lot. The same appreciation was not extended to my IT department, which had decided -without informing me- that the service tool I had installed prior to departure was a potential risk and had quietly removed it...

Besides, if you are less -ahem- cost-sensitive, Inmarsat offers FleetBroadband which from geostationary orbit easily gives you >100kbps (as long as you are not anywhere near the poles!)

As an added bonus, rain fade is not much of an issue; I believe it uses L band. (Same as GPS/GLONASS/Galileo)

Sounds like we could swap stories. :-)

Trying to one text messages in and one out was a 20 minute affair in the rainy mountains of Peru with Iridium. Globalstar data service seemed to have random drops and periods where connection was impossible, even in dry weather.

FleetBroadband sounds interesting.

You still have limited bandwidth and latency. Just because you break the distance up into smaller hops, the total distance to travel remains the same. Maybe scientists should focus on increasing the speed of light. LEOs also have short orbital lifetimes meaning the constellation needs to constantly be refreshed which isn't cheap.

The total distance to travel going via GSO or LEO is not at all the same; in particular, there's a fairly wide range of positions that an LEO satellite can be in where it has LOS to both Hawaii and the mainland.

That being said, once you've laid the cable, fiber optic is fast, they typically over-provide the cables when they lay them, and they can often increase capacity by improvements at the endpoints. Radio to LEO seems unlikely to compete with that, but for less-populous and/or poorer islands it might make sense.

I think SpaceX proposed system will be in LEO ~1100km above the earth. At this distance, satellite internet should be faster due to having a straight line, and the fact that EM in a vacuum is faster than fiber.

How about a chain of microwave point to point buoys leading to the mainland?

Troposcatter words for longer range communications - its better than LOS.

It doesn't seem that primitive to me. You already have connections for other utilities. Even in my home, I prefer to run Ethernet rather than deal with Wifi issues to various devices.

You have to step back from the whole concept surrounding wireless communications, and really weigh the fundamental principles at work to consider the realities of the universe we inhabit, which is totally weird and doesn't at all stand up to ideas one might consider "normal" or reasonable.

The idea of wireless communication is only just a little bit more glamorous, because it's really the only thing that will assist humanity as part of space travel. And because we know light signals have pretty high fidelity across vast distances of (mostly) empty space.

But, solid state conductors, even if not purely contiguous due to repeaters and amplifiers, still seem pretty rational, when you start to think about what conductive transmission mediums actually do, especially as compared to emission filaments and diodes as broadcast signal sources.

You're close to why point to point connections have an advantage, well two.

Fixed links mean the signal can be narrowed so that less other things interfere with it and it interferes less with other things. There isn't contention or the crowded room effect (like an apartment complex stuffed with WiFi nodes).

The fixed nature also allows for a more optimal end point configuration. (Lower power, longer distance, etc)

Yep, right out of the box, the clear advantage is no deliberate signal jamming with a malicious high volume noise floor, no line of sight requirements for directional signals, and lower incidental destructive interference from sharing the medium, reducing dependence on digital signal processing filters.

If anyone is interested in making a fancier version of this, it would be very cool to add infrastructure on land, too. EG, Level 3 has a map at http://news.level3.com/network-map, or Allstream's at http://www.allstream.com/about-us/ipnetwork/

I've always thought it would be cool to use large area traceroutes and BGP dumps to automatically infer some topology data too, automatically creating such a map.

Trouble with trying to map traceroutes is so many long distance runs are in mpls that you can get 1 hop from Europe to Japan!

This seems very similar to https://www.submarinecablemap.com/.

For the cables I'm aware of, the one you linked seems a little more accurate. On Greg's, there's a pin on the map but it just says "0 cables land here"

I found this to be a fun reference when reading Cryptonomicon by Neal Stephenson.

No mention of submarine cables is complete without reference to Neal Stephenson's excellent feature for Wired [0] - 'Mother Earth, Mother Board' in which he tags along as an intercontinental cable is laid down.

Also, Discovery Networks did an episode of its 'Mighty Ships' series on a Tyco Reliance class cable ship - quite interesting if you can find it, although a bit short on nerdy detail, as one would expect.)

[0] https://www.wired.com/1996/12/ffglass/

The site looks down. Probably because womble.ddns.info is down:

dig result:

    ;; ANSWER SECTION:
    cablemap.info.          578     IN      CNAME   womble.ddns.info.
    womble.ddns.info.       29      IN      A       41.180.82.42

Archive.org link from 2017-08-30:

https://web.archive.org/web/20170830051926/http://cablemap.i...

Also Google Cache http://webcache.googleusercontent.com/search?q=cache:_BoKori...

Looking at the map I suspect it's still the case a good many of the these cables end up trunking into the same building / using the same power and can be described as very much being in a "shared risk group".

You think you're buying a backup link resilience and you're paying lots of money to a different carrier, but one proverbial backhoe or one block looses power and oops you end up with a network partition.

Interesting that there are no current transpacific cables at all landing in western Canada. Makes the area very dependent on the neighbour to the south (which is an earthquake risk, as well as a political risk).

Ehh, I imagine if things get too bad they can always lay a new one. In the meantime Western Canada could probably be connected to Eastern Canada somehow. Given that people were laying tons of cables 120 years ago, it can't be _that_ difficult for a country as big as Canada in crisis.

Canada is well connected internally with Fiber

There are overland fiber runs. One map is:

http://www.allstream.com/about-us/ipnetwork/

Those will connect with arctic and atlantic undersea cables, so Canada wouldn't be completely cut off.

which is an earthquake risk, as well as a political risk

Oh, come on. Maybe an earthquake risk. But, really, "political risk"? The longest undefended border in the world?

Reply on news.ycombinator.com