[t3rmi]

Thanks. It looks like their orbit is around 800 miles compared to the usual 22,000. Expected latency seems to be around 25ms. Would satellites be used for first hop only? I image latency would quickly increase if used for multiple hops.

[sjwright]

I imagine SpaceX could eventually perform the majority of routing in orbit, with packets returning to earth for the "last mile".

If they're successful, you could potentially see a high bandwidth SpaceX link on the roof of every major data center on the planet. They could be self-reliant for connections to every major online service.

If they're very successful, you could potentially see data providers like Netflix and CloudFlare joining SpaceX in orbit, most likely embedded into a future SpaceX satellite...

[tbabb]

Space is just about the worst place to put a datacenter.

Price to orbit is thousands of dollars per kilo. Maintainability is zilch. And you have lots of cosmic rays flinging charged particles into your sensitive memory banks, flipping random bits. Also, heat management is very difficult.

I couldn't think of a worse place to put a datacenter than space. The bottom of the ocean would make more sense.

[mtanski]

Netflix ISP CDN in space. Now that'd be something seriously cool.

[jessaustin]

It's funny; Netflix is one provider that doesn't need low latency so they would be fine with more traditional satellite orbits.

[barkerja]

It wouldn't be a wise decision; what if they one day decide to enter into live programming (sports, news, etc.), where latency does matter?

[jessaustin]

Latency doesn't really matter for that either. Seeing a football play 5 seconds too late doesn't hurt anyone.

[imtringued]

And existing streaming services already have a latency of several seconds. Twitch for example can have a delay from 5 to 20 seconds or more. Unless you're using WebRTC or similar technologies you won't see latencies below a second.

[NullPrefix]

>Seeing a football play 5 seconds too late doesn't hurt anyone. Only if the audience do not know about it.

[agildehaus]

Once the constellation is built, you can do worldwide routing in orbit, which is actually more efficient latency-wise (at least for international hops) as you're not dealing with political and physical boundaries and light is passing through vacuum not fiber.

[jandrese]

While this is possible, it is very challenging to actually achieve. Your latency numbers go to hell in a hurry if you're bouncing around satellites in orbit (eating a queuing delay on every hop). Iridium did this, but at the end of the day their mobile to mobile latency numbers are similar to Geo satellites.

SpaceX's plan is almost absurdly ambitious and there is a lot of skepticism in the comm satellite community that they will achieve their stated goals, especially at an affordable price point.

[mygo]

> especially at an affordable price point.

They don’t necessarily have to be affordable. You can be premium. They just have to be able to provide high speed internet and not be Comcast.

They might even make it realistic for more people to live out in rural areas where there either is no internet or what’s available is inhibitingly slow. That would be an interesting niche. People who choose to live out further than where telecomm companies will support have money to spend too. There’s pent up demand.

[lsaferite]

>there is a lot of skepticism in the comm satellite community that they will achieve their stated goals, especially at an affordable price point.

Not that I have any experience in that industry, but didn't the commercial rocket launch have the same set of opinions?