The Dunant subsea cable • Hacker News Board

Re: The Dunant subsea cable

#171

Post by ksec » Thu, Feb 04, 2021, 6:18 PM UTC

sneak wrote: post #141

ksec wrote: post #52

I know a lot is focusing on the Bandwidth. But are we making any progress on Long Distance Subsea Cable using Hollow Core Cable, achieving close to maximum speed of light for theoretical lowest latency possible? Imagine cutting latency from West Coast US to Hong Kong by 50ms! Light is only traveling at around 2/3 of speed within Fibre. The previous decades have been around Bandwidth. Is time we shift out focus to lat…

I agree, but we can start on our local machines first. Most of the latency of modern computers isn't related to the network.

Yes. Keyboard, Mousepad, Display, Sound, Graphics.

I mean input lag [1] is easily 50ms. But some of them requires software changes. And any thing software is expensive. The cost of this new Cable is only $300M. Hardware innovation is getting faster and cheaper than Software.

[1] https://danluu.com/input-lag/

Re: The Dunant subsea cable

#172

Post by potiuper » Thu, Feb 04, 2021, 6:18 PM UTC

rorykoehler wrote: post #142

idlewords wrote: post #58

Earlier quoted context omitted.

You gotta bore for that sweet latency win. A chord tunnel between San Francisco and Hong Kong would save 1300 miles (20% improvement right there), and if you drill it straight enough, you won't even need a cable.

How deep would this bore tunnel be at the centre most point if it was perfectly straight? Would it go into the mantle?

d=r(1/cos(s/(2r))-1)=3958(1/cos(6906/7916)-1)~=2198 mi; Yes, into the lower mantle with only 692 mi to go to the outer core. What I would love the internetz to explain is how to justify the h8 for HFT, yet the luv for Musk since he is the one in the driver's seat at the moment for this stuff with StarLink and the Boring company.

Re: The Dunant subsea cable

#173

Post by faitswulff » Thu, Feb 04, 2021, 6:18 PM UTC

nippoo wrote: post #119

This is super-cool! I found "enough to transmit the entire digitized Library of Congress three times every second" to be a really weird comparison though - I'm used to text being really small and compressible, and I doubt many people have an intuitive grasp of how much One Scanned Library of Congress is. How many hour-long Netflix/YouTube episodes per second, on the other hand...

As of March 2019, Netflix is reported [0] to have 60 petabytes of data. Google tells me that 60 petabytes / 250 terabits per second comes out to 32 minutes. I’m not sure that translates to the layperson who might not appreciate what a petabyte is, but in the space of a single show you could theoretically transfer the contents of Netflix’s entire library over this pipe. So basically almost enough bandwidth to transfer the average user’s porn stash in a single day!

[0]: https://zeenea.com/metacat-netflix-makes-their-big-data-acce...

Re: The Dunant subsea cable

#174

Post by nameless912 » Thu, Feb 04, 2021, 6:20 PM UTC

easton wrote: post #42

ed25519FUUU wrote: post #40

Earlier quoted context omitted.

Google uses gcp.

For everything? I was under the impression they still ran all the big stuff on their internal cloud with Borg and all the other infrastructure tooling they built.

My understanding is that GCP is essentially selling off extra capacity in those data centers, so for example your VM running in GCP is scheduled by Borg under the hood. So it's more like GCP runs on Google, rather than Google running on GCP.

Re: The Dunant subsea cable

#175

Post by ptudan » Thu, Feb 04, 2021, 6:23 PM UTC

mrtnmcc wrote: post #169

mmmBacon wrote: post #164

Earlier quoted context omitted.

Starlink satellites are in orbit 550km high. So any journey would add at least 1100km. Moreover not sure that a single satellite would be able to hit another one across transpacific distances and may need to go through multiple hops to get there. Each hop will add latency since signal needs regeneration. So it’s not clear to me a swarm of satellites is a real winner from a latency POV. Furthermore, given costs to put…

>> Starlink satellites are in orbit 550km high. So any journey would add at least 1100km Might want to check with Pythagoras on that one..

Meh, he said at least. There could be cases where you beam up then down nearly vertically (same city).

Re: The Dunant subsea cable

#176

Post by tinus_hn » Thu, Feb 04, 2021, 6:24 PM UTC

john37386 wrote: post #112

lifeisstillgood wrote: post #107

Old timer story - years ago, Demon Internet (my old employer) was beginning to enter super-growth and wanted to really set itself up for trans-atlantic connections. So they decided to buy a T1 - 45Mbps link across the atlantic. Now it turns out that BT had only ever resold fractions of T1s - they had never actually had anyone want a whole one. And as such their sales commissions did not cap out. So, we rang up, a sal…

T1 maximum speed 1.544 Mbps T3 is about ~45 Mbps Reference: https://en.wikipedia.org/wiki/T-carrier

In internetworking a Tier 1 carrier is a carrier that is so interconnected other parties pay to receive traffic from them.

Re: The Dunant subsea cable

#177

Post by mrtnmcc » Thu, Feb 04, 2021, 6:29 PM UTC

0xffff2 wrote: post #65

m_eiman wrote: post #57

Earlier quoted context omitted.

Won't they be at low enough altitude that they'll need more hops than fiber to get around the globe, where each hop adds at least some delay?

When you're traveling at the full speed of light in vacuum, compared to 2/3rds in fiber, even a few extra hops can leave you with significantly lower latency.

Right, if they are using standard OTN framing, the hop latency should be ~3 microsecond (which is <1km of light propogation)

Re: The Dunant subsea cable

#178

Post by function_seven » Thu, Feb 04, 2021, 6:34 PM UTC

ptudan wrote: post #175

mrtnmcc wrote: post #169

Earlier quoted context omitted.

>> Starlink satellites are in orbit 550km high. So any journey would add at least 1100km Might want to check with Pythagoras on that one..

Meh, he said at least. There could be cases where you beam up then down nearly vertically (same city).

So, "at most" then, right?

The further you are from the other end, the less additional distance the satellite adds on.

Re: The Dunant subsea cable

#179

Post by cameldrv » Thu, Feb 04, 2021, 6:43 PM UTC

jxcl wrote: post #127

thomaslangston wrote: post #122

Earlier quoted context omitted.

More feasible would be transmitting neutrinos or some other signal that would not be blocked by the Earth.

> More feasible If they don't interact with the thousands of miles of earth between the source and the destination, they probably also won't interact with the receiver! :p Imagine the retransmission rates! https://en.wikipedia.org/wiki/Neutrino_detector

There has been a lot of progress in the past 20 years on antineutrino detection. Antineutrinos are produced by fission and so there's been a fair bit of interest in detecting them to detect covert nuclear tests as well as potentially a new modality of detecting nuclear submarines.

I think it could become possible before too long to use this to transmit data. It would probably be a ~billion dollar project, but the HFT arbitrage market is essentially winner-take-all, and may be large enough to support this size investment.

Re: The Dunant subsea cable

#180

Post by ceejayoz » Thu, Feb 04, 2021, 6:51 PM UTC

BelenusMordred wrote: post #135

ceejayoz wrote: post #61

Earlier quoted context omitted.

This was done in the 70s/80s, but I doubt it's worth the effort now. It only worked because the Soviets assumed the cables were inaccessible. End-to-end encryption is a thing even for the general public now. Now we just compromise the servers/routers. https://gizmodo.com/the-nsa-actually-intercepted-packages-to...

> I doubt it's worth the effort now It's very much still happening. Metadata is enough for intel purposes, storage is ridiculously cheap and post-quantum breaks of key exchange is forever 20 years away like fusion. https://www.nytimes.com/2015/10/26/world/europe/russian-pres... https://www.theatlantic.com/international/archive/2013/07/th... https://www.zdnet.com/article/spy-agency-taps-into-undersea-...

The first link describes attacking cables - severing internet access for entire continents.

The second is from 2013; Google and others encrypted those comms shortly afterwards after Snowden revealed those taps. https://arstechnica.com/information-technology/2013/11/googl...

> "The traffic shown in the slides is now all encrypted and the work the NSA/GCHQ staff did on understanding it, ruined."

The third link is twenty years old, and no longer very doable for the same reasons as above. Anyone still sending unencrypted stuff along these cables deserves to get stung.