Starlink network topology simulations and predictions | Hacker News Reader

Starlink network topology simulations and predictions

reddit.com

34 pointsmastax8 comments

I guess the elephant in the room is spatially uneven bandwidth demand. How many Netflix 4k streams will the constellation handle if a significant portion of London residents switches to Starlink? The simulation says a few dozen sats visible at any moment; will they handle a few terabits that I think London might saturate?

Why would you use this instead of 4G if you are in a city like London though? It will surely be more expensive.

>How many Netflix 4k streams will the constellation handle if a significant portion of London residents switches to Starlink?

I wonder if as access to orbit cheapens how long it will be until we start getting orbital datacenters/CDNs specifically to mitigate this type of issue.

How would you set something like that up, orbitally speaking? Wouldn't the satellites move past <insert city here> pretty quickly?

I imagine an orbital server farm would probably be in a higher orbit than the Starlink satellites, and would just connect to them, vs people planetside connecting directly.

Starlink sats already move past <planetside location> quickly, which is why they propose loads of satellites and phased array antennas that switch from satellite to satellite as they move by.

How does that provide an advantage over having the same data-center on the ground? The starlink satellite still needs to communicate with both downstream clients and some kind of upstream link (whether that link is geographically above or below it doesn't seem to provide any obvious advantages for the starlink satellite itself, and the advantages to having the datacenter on the ground are numerous).

Unless they somehow direct waste heat efficiently to maintain the orbit I can't see how you'd put enough up there that's useful.

There's a finite amount of bandwidth you could point down, you could use all that serving customers and link to another satellite not in that signal path and relay through that to a datacenter for a small rtt increase.

I bet there's some clever way you could use rtt from leo to geo orbit as a storage medium itself

StarLink will end up pricing out that kind of demand in populated areas. It will be useful for traffic that needs low latency, and the people who need that will price others in the area out of StarLink services.

So the surgeon conducting teleoperated surgery will be able to price out the gamer who just wants CS:GO to be a few milliseconds more responsive.

Reply on news.ycombinator.com