An underwater landslide knocked out two critical submarine cables serving Africa | Hacker News Reader

An underwater landslide knocked out two critical submarine cables serving Africa

kentik.com

118 pointsdtjb35 comments

Interesting that latency between Cape Town and Seoul actually improved (decreased) as a result of the cable cut, due to traffic being allowed on a more direct (and more expensive) route between the 2 cities.

More examples in Slide 10 of https://www.linkedin.com/in/dougmadory/details/featured/5001...

Here : https://en.m.wikipedia.org/wiki/Braess%27s_paradox

This is the principle behind it I believe.

It's a reaaaally cool rabbit hole.

I don't think that's the issue here if the poster is correct that the better route is more expensive than the one that failed.

It's not irrational or paradoxical to use a route that's less expensive regardless of quality, but fail over to a more expensive route that also happens to be better quality.

If you were planning to send something in the mail because it's low cost, but the post office was closed, you might use a courier service instead, pay more, and get your item to the destination quicker. No paradox there.

Deep linking into apps on mobile has become such a pain of late.

I think the content that was linked is the slide deck in this blog post from the same author: https://www.kentik.com/blog/dual-subsea-cable-cuts-disrupt-a...

Interesting - the service my team runs has had chronic latency problems with South Africa, so I wonder if I can see a difference in the metrics.

I thought that there was a routing algorithm that was supposed to optimize this?

Routing algorithms are tunable. Especially BGP, which is what is used to route between internet carrier companies, technically known as "autonomous systems".

BGP by default prefers the path going through the least number of autonomous systems. A carrier will announce the networks they can reach. Each carrier handles routing within their autonomous system separately, with other routing protocols. BGP is concerned with routing between autonomous systems, not with routing inside each system. It is common for BGP carrier implementations to set certain paths as preferred based on contract terms with other carriers, i.e. cost.

Other routing protocols work differently. Some of them by default prefer paths with the most bandwidth. Latency is usually correlated with hop count and bandwidth. The devil is in the details.

I'd imagine it is setup to account for route cost and latency.

404

Same, the link's down.

Nothing interesting to add, just an anecdote. As someone based in a semi-rural part of South Africa, this dual cable break was a huge PITA for a few days. Latencies went through the roof; remote work was made very difficult.

Fortunately work not noticeably affected, but Spotify can’t manage to stream, I have to use downloaded music or it constantly stops.

My web client was behaving waaay better than my desktop client on this issue.

I read "submarine cables" and imagined a sort of monorail or guide-wire for an underwater vehicle. I thought, "I guess that makes sense, that'd probably be cheaper or safer than an untethered submarine".

It's funny how "good" the brain is at rationalizing misconceptions like that.

If I hadn't read a couple comments to try and see why Africa (or anyone) would need a safety-rope-assisted underwater submarine I might have gone on assuming that was just a normal thing.

Due to some inane regulations there is only one ship that can fix the cable break and it was busy over 8000 km away fixing another cable break - will take a week to get onsite.

Cloudflare Radar has a decent visualization of the event.

https://radar.cloudflare.com/cg?dateRange=14d

> According to him, one of the cables had been dragged "over a kilometer"

Undersea landslides can go for over a hundred kilometers and can last hours.

So I'm not sure why this is impressive or in question?

It doesn't seem unlikely to me.

Good journalists cite their factual claims, even/especially if they're believable.

I've never heard of underwater landslides, and I don't think it's a reasonable assumption that a reader knows anything about them.

> I've never heard of underwater landslides

They are a huge cause of tsunamis, with the slips often precipitated by earthquakes.

In geology, there's a literature about cable breaks caused by "turbidity currents".

In particular, this 1952 paper by Heezen and Ewing is celebrated https://gwern.net/doc/technology/1952-heezen.pdf

> I've never heard of underwater landslides

Good heavens. You do surprise me. There have been some vast ones, e.g.

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

Some have the potential to massively affect human life well above sea level, e.g. La Palma in the Canary Islands, which when it goes will inundate much of the east coast of the USA:

https://www.e-education.psu.edu/earth107/node/1609

The video from that is sensationalist but makes some good points:

https://www.youtube.com/watch?v=6utAunBKXV4

Facts are sometimes just facts, not impressive fun facts.

>It doesn't seem unlikely to me.

It's not unlikely, it's probably true.

But it's at least a little coincidental that it happened during a coup in Niger.

"A communications disruption could mean only one thing… Invasion." —Sio Bibble

Reply on news.ycombinator.com