Personal bugbear: Ireland has a lot of connectivity but only three cables (two from Cork, 1 from the north-west) go to a destination other than the UK. All other Irish internet/telephony traffic goes through the UK. Remember that there are many European data centres and EMEA headquarters in Dublin.
Much of it goes through Bude in Cornwall, which has a GCHQ listening post perched over the submarine cable (https://en.wikipedia.org/wiki/GCHQ_Bude). Or through Southport which is also known to be monitored by GCHQ.
I find it surprising, particularly as the UK prepares to leave the EU, that this issue isn't of more concern
MITM attacks are something that only a handful of governments worry about. Everyone is still more concerned about unpatched libraries, SQL injection attacks, and viruses.
Pretty interesting. I'd be cool to have one that showed the cables on land too. I wonder if there's a tool like that, be useful to plan where the best connectively is if you are building your own datacenter for example.
If you're building your own datacenter (and by this I define a serious project with a budget $20m+), you're going to be hiring a person as a full time staff member who knows where all of the actual OSI layer 1 and 2 operators have infrastructure in your local state/province. For example the persons who are employed by Sabey related to their huge datacenter operations in Quincy, WA.
Much of who actually 'owns' fiber in a region is opaque for competition reasons. You have the company that actually laid the cable, then you have companies which may have purchased a few dark strands or IRUs in the cable, then you have a myriad of companies that are leasing dark and installing their own DWDM systems.
Where you can get access to dark also depends on geography and opaque business relationships. There's a lot of places in the US where a 288-count fiber cable goes through but here's no regen hut or POP built to access it. You need you build backwards or forwards along the path to get to the nearest place with active electronic equipment.
I was just thinking the same thing. Not just for datacenter planning, mind you, but even for a neat representation of which cities are more likely to have high-speed Internet connectivity; there are probably significant differences in expected speeds and costs between some city way out in the middle of nowhere and some city sitting right on top of a major data artery.
It shows some overland links as well but it doesn't show intermediate hops. For example, you can see from other maps that most Pacific cables stop off in Hawaii.
You can match the nodes to entries in a trace root log. I did this for a chapter (on network latency and performance) in my last book.
Overland links to Europe are most of it. If you look at the number of major Russian ISPs that are members of the DE-CIX in Frankfurt, that's a good indicator of traffic flows.
The lack of land connections on the map threw me for a loop, too; it makes it look like there are no direct connections between North America and Iceland (or Greenland, for that matter) due to the apparent gaps where there are (hopefully) overland cables.
My first guess would've been to support a connection between North America and Russia, but there doesn't seem to be such a connection on the map.
My second guess is that there might really be a significant data demand there. Between research stations, defense stations (gotta protect 'Murica from those damn pinko Soviets), oil drilling, and at least one or two decently-sized cities, the need for those cables doesn't seem unreasonable.
EDIT: the one covering the northwest Alaskan coastline appears to be the first phase of a cable connection through the Northwest Passage.
1. Satellites aren’t used because they can’t carry terabytes of data for less than a billion dollars per communication line.
2. The bandwidth available using a single fiber optic cable and a laser beam is much much greater than you can get from a single satellite radio channel. This is due to the higher frequency and shorter wavelength of light compared to microwaves. The higher the frequency, the greater the bandwidth.
3. An undersea cable is a bundle many fiber optic cables. Consider each fiber cable as a channel. You can have more channels, each with a higher capacity, than you can build radio channels into a satellite.
4. The uplinks and downlinks cost and putting the satellite in space is a huge huge ask and far more risky.
5. The delay for satellite communications would be around 255ms both uplink and downlink. For continuous traffic this not to a bad price to pay. But for burst traffic (like voice) you pay for the delay at each pause. The Rule of Thumb is 10MS per 1000 miles so Rule of Thumb to Europe on say TAT-8 would be about 75MS vs 510MS for satellite.
6. Finally, you can fix a broken cable. Once you launch the satellite you don’t get a chance to fix it if it gets broke.
re: #3, not exatcly, an underseas cable is actually relatively few strands compared to a terrestrial cable. Many older ones are only two strands. Modern ones 4 or 6 strands. It's the DWDM terminals on each end of the cable that enable fun things like 80 channels x 100 Gbps full duplex coherent QPSK per channel.
Latency, cost, and connection quality are usually the three big issues with satellite connections. For those reasons, they're usually only viable in remote areas with low populations (or otherwise low need for electronic communication); at some point, a physical connection ends up becoming necessary.
1) Satellites don't use fiber as a medium so have that against them
2) from point A to a satellite back to point B is generally going to be a pretty significant distance, as such, when combined when a comparatively slower medium you'll get increased latency, which is bad for interactive applications.
3) satellites can be susceptible to weather issues.
To clarify/correct here: satellites do not use a slower medium than fiber; satellite (and radio in general) and fiber communication both happen at the speed of light (since they both use light to communicate). If we had a space elevator, and someone were to string a fiber-optic cable along its length and ping something at the other end of it, it'd take just as long for that signal to reach the top of the elevator as it would to reach a satellite orbiting right next to it.
The "slowness" is due to the other factors you've described (the longer distances and the increased risk of environmental interference).
a geostationary satellite that weighs 6000 kilograms and costs $185 million to build and launch has less aggregate data throughput capacity than two strands of singlemode fiber in a cable with a diameter thinner than a pencil. And that is before you get into the fact that medium/long distance fiber these days is generally built with a cable that has a minimum of 144 strands in it, if not 288 or 864.
It's not even close when you're talking about 40 or 80-channel x 100 Gbps DWDM systems vs. satellite, whether it's geostationary or o3b.
Satellite is best used to reach very difficult to reach locations (example: Skardu Pakistan) or for maritime purposes.
Much as this comment is like 40% trolling and likely to lose me what little karma I have, is there an easy way to see this data in a better projection?
Much of it goes through Bude in Cornwall, which has a GCHQ listening post perched over the submarine cable (https://en.wikipedia.org/wiki/GCHQ_Bude). Or through Southport which is also known to be monitored by GCHQ.
I find it surprising, particularly as the UK prepares to leave the EU, that this issue isn't of more concern