Rough translation of the article in Phys Rev E.

Some parts of the Internet are very highly connected, and some not so much. If you want to get from one badly connected place to another, you probably have to through one of the very connected nodes. This makes the distance function on the network look hyperbolic. (This sounds reasonable.)

You can make routing easier on the Internet by knowing this, and so only looking for hyperbolic paths (or geodesics) when trying to get from one place of another. All seems pretty fine by me!

topographical?

Is that initial quote in para 2 right? A topographical map does take account of geometric coordinates, a *topological* one doesn't.

No surprises abound.

Networks support almost no-one. Static addressing needs simple architectures. Nevertheless, shockingly and notwithstanding such advances, nobody sees a node system application here.

This idea has zero chance of being implemented I suspect.

title

Closes neighbour geographically is an almost meaningless metric, since there are so many ways two routers could be connected, different loads, etc.

What you need is closest by latency, which the existing least cost routing protocols do extremely well. It only looks chaotic to our eyes because we can't take into account all the variables.

Routing table size is a problem (there are solutions, and ipv6 helps a bit) but the internet won't 'collapse' just routers will get bigger/faster.. just like everything else.

Oh the irony

Caída means "fall" in Spanish; and it is usually used to refer to system outages. "Se cayó mi internet" (my internet went down), "se cayó el servidor" (the sever went down), "el sistema está caído" (the system is down) ... it does bring some chuckles to read that acronym.

About the hyperbolic routing ... all this time I thought that the real problem with the internet was the lack of redundant links, such that a boat slashing an underwater fiber cable will leave a whole continent offline. Whoops!

Well but with IPv6

With IPv6 we will have a lot less route fragmentation.

Well....

I may be drunk but from my poor and unenlightened understanding of how the internet thingy works, this sounds all like a pile of ballocks.

No, really, do these guys actually have any concept of routing or how the internet works? I'm pretty sure the interwebnetthingy can grow and grow with its current model and survive just nice and dandy.

tree falling in the woods

OK so a couple of guys measure the reachability of internet endpoints from other endpoints. If some obscure site becomes unreachable it ticks their black-hole count but most of the internet doesn't even notice.

I'd speculate that traffic within a country can be routable and most of the traffic is self-contained within said country. Only when you have to start crossing the boundaries does it get complicated. But then how many people in the US (as an example) would notice if something in Lower East Anywhere (pick a country, any one of them) fell over? Not many.

easy fix

Block facebook

Just saying, but...

I've been hearing this one since middle school, well over 10 years ago. That doesn't mean it's not true now, but it's not like they were right back then.

"Google will be trying to stay on the center of said sphere"

either that or a burning TARDIS.

I know which I'd prefer.

(P.S. Any chance of seeing that 'hyperbolic map' being carried through space on the backs of, say, four enormous elephants standing on a turtle?)

Setting things straight

First : I have no doubt that there are very intelligent people who are indeed capable of pinpointing an operational issue in the functioning of the Internet. As such, given that these people know what they are talking about, there is bound to be a certain amount of problems that can arise in our daily use of this increasingly-indispensable tool.

Second : Given the amount of money generated by the Internet in all of its commercial activity and interests, I have no doubt either that this issue will be solved in one way or another without impacting my browser in any way.

Third : this is already not the first time an Internet meltdown has been forecast and, gosh, it's still there isn't it ?

Conclusion : this is not an issue for us normal people - alert condition martini dry, thank you for your attention.

There's an even simpler solution...

Block YouTube and a few of the 'Free Erotic Chat' sites.

That'll lower the traffic enough that there's no problem anymore.

And if that's not enough, take out a few torrent sites and p2p networks.

Seems VoIP centric to me.

I guess it depends heavily on the type of traffic, many of the major aggregators here appear to be voice carriers, thus the graph if biased towards VoIP traffic. Pobbily because that's who requested / paid for the analysis?