3 ms·

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 (

by cj 3y ago

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/50016397/single-media-viewer/?profileId=ACoAAABgDfMBQqg6K3WmEobvLYAvoetjdTSF-R0 https://www.linkedin.com/in/dougmadory/details/featured/5001...

bruce343434 3y ago
404
- MichaelZuo 3y ago
  Same, the link's down.
13of40 3y ago
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.
dmbche 3y ago
Here : https://en.m.wikipedia.org/wiki/Braess%27s_paradox https://en.m.wikipedia.org/wiki/Braess%27s_paradox This is the principle behind it I believe. It's a reaaaally cool rabbit hole.
- toast0 3y ago
  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.
jasfi 3y ago
I thought that there was a routing algorithm that was supposed to optimize this?
- mikem170 3y ago
  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.
- Denote6737 3y ago
  I'd imagine it is setup to account for route cost and latency.
jbergler 3y ago
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-african-internet/ https://www.kentik.com/blog/dual-subsea-cable-cuts-disrupt-a...

by cj 3y ago

bruce343434 3y ago

404

MichaelZuo 3y ago
Same, the link's down.

13of40 3y ago

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.

dmbche 3y ago

Here : https://en.m.wikipedia.org/wiki/Braess%27s_paradox https://en.m.wikipedia.org/wiki/Braess%27s_paradox This is the principle behind it I believe. It's a reaaaally cool rabbit hole.

toast0 3y ago
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.

jasfi 3y ago

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

mikem170 3y ago
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.
Denote6737 3y ago
I'd imagine it is setup to account for route cost and latency.

jbergler 3y ago

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-african-internet/ https://www.kentik.com/blog/dual-subsea-cable-cuts-disrupt-a...