All your criticism is valid in the long run but right now, SpaceX'es focus is to scale up, focus on usability for majority of its customers and become profitable. Removing any unnecessary feature is a must in order to reduce risk.
back
3 comments
> SpaceX'es focus is to scale up
> Removing any unnecessary feature is a must in order to reduce risk.
You can’t scale a space based planetary ISP without IPv6, this isn’t a feature, it’s a requirement.
IPv6 amounts to a firmware update which the current hardware is and any future hardware will be capable of. The major hurdle in scaling Starlink is fast and cheap deployment of 1000s of satellite hardware. At the moment, the sole focus of SpaceX in relation to Starlink is to get their V2 satellites to orbit in order to keep up with bandwidth demands. V2 requires the Starship system which is yet to make it to orbit.
TL;DR bigger fishes to fry ATM - yes you need IPv6 to scale, no you don't need it right now.
It’s really too bad that ipv6 is only… checks notes… 26 years old now. I realize that may be an unreliable metric, so it’s roughly equivalent to 2.88 react.js lifetimes, or 3.25 vue.js lifetimes.
When the ipv6 spec was released, the latest python release did not yet support list comprehensions.
In other words, there is no reason to not support ipv6 out of the box in 2022.
> In other words, there is no reason to not support ipv6 out of the box in 2022.
i'd go even further and say that no ipv6 support means obsolescence.
> there is no reason to not support ipv6 out of the box in 2022
Use the age of a service as the metric, not the absolute year we're in. It's probably reasonable to say that there is little reason not to support IPv6 for an ISP with X years of operation. Starlink is young still.
An analogy is worldwide sales for a new laptop company. You can say that in the age of globalism, there is no reason not to ship to every continent right off the bat. But for a startup with limited cash that has lot of building blocks to lay out, it's a huge risk. They should plan for it, but only branch out when they've got a solid foundation.
A better analogy would be a new laptop company selling laptops with Windows XP.
"Our staff is more familiar with XP. We promise we're still working on the Windows 10 drivers, but in the meantime you can try to run it in VMWare."
A new ISP should implement IPv6 first and then run IPv4 on top of it like T-Mobile. They shouldn't "add" IPv6 because it should have been designed in from the beginning.
They already had IPv6 support while they were still using google cloud for connectivity (not sure why they went with google for their initial phase instead of a more traditional carrier) and when they moved to their own network they disabled IPv6 for some reason.
The funniest bit is that I'm not sure you can get normal v6 on GCP yet...
> no you don't need it right now.
Yes we do. IPv4 exhaustion is a thing.
That's what CGNAT provides a temporary solution for.
I doubt there's much overlap between the people working on IPv6 and those working on getting Starship off the ground.
It's not a function of overlap. If they've determined IPv6 isn't a priority and instead a risk, then it makes no sense to dedicate resources to it right now. It's not as if everything else about the firmware/software is wrapped up and the software team is sitting on their hands doing nothing.
CGNAT itself is a enormous risk. It's WAAAAAAAAAAAAAAY more complex and unstable than v6.
> CGNAT itself is a enormous risk.
How?
IPv6 isn't a substitution for CGNAT, it's an addition to it. You either have to keep CGNAT or replace it with dedicated IPv4 for each customer. Dedicated IPv4 is most likely too costly given the limited availability. SpaceX is also trying to cut cost aggressively.
It’s stateful.
Specifically, if a plain router stops working, BGP will route around it, and all you did was drop packets. If NAT stops working, you don’t just drop packets, you drop whole connections. Applications don’t tend to tolerate dropped connections as well as they tolerate dropped packets.
True. As an ISP you are gonna need an IPv4 stack no matter what. Even if that stack is CGNAT'd up the ass. I can't even ping news.ycombinator.com or amazon.com with IPv6.
That is the biggest problem with IPv6. Who is gonna be the first ISP to shut off their IPv4 stack? There is always gonna be some random website that is IPv4.
When all the big services become IPv6, the number of IPv4 megabits will become small.
You might just direct all the v4 traffic via a tunnel to another ISP which specializes in legacy services like IPv4, running SMTP/news servers, etc. Now you've saved all the cost of maintaining all the IPv4 peerings and config.
>When all the big services become IPv6, the number of IPv4 megabits will become small.
Well I have been hearing about the end of IPv4, and IP exhaustion for about 20 years now, and I fully expect people to still being moaning about it 20+ years from now while the majority of the interment still communicates over ipv4
Amen. I dual stacked my home network 10 years ago. 5 years ago I joined an ISP that gave me CGNATv4 and IPv6 and I opted to disable IPv6 at the router.
Why? If you have an IPv4 address, even a dynamic one, then IPv6 may not offer you a lot of practical benefit, but CGNAT-only sucks if you're at all technical.
It's a partial substitution. What percent of your traffic is youtube, for example?
Where did you get that information?
From Elon Musk in numerous interviews. Latest of which is with Tim Dodd.