Amazon Project Kuiper Broadband Constellation

[Tomness]

If data is going from one side of world to other then packets will pass through multiple international routers/exchanges.

Let's clarify that: "... packets will pass through multiple international routers/exchanges owned and operated by different companies-governments-collectives-whatever". That's the way the Internet works. In that context, Kuiper will be just another carrier among many. Kuiper may provide something special to their subscribers, or on behalf of there subscribers. So what is that "something special"?

What Google and Starlink promised but yet to deliver. Terminal to Sat to nearest Data Center. Only 2 hops...

[mn]

If data is going from one side of world to other then packets will pass through multiple international routers/exchanges.

Let's clarify that: "... packets will pass through multiple international routers/exchanges owned and operated by different companies-governments-collectives-whatever". That's the way the Internet works. In that context, Kuiper will be just another carrier among many. Kuiper may provide something special to their subscribers, or on behalf of there subscribers. So what is that "something special"?

What Google and Starlink promised but yet to deliver. Terminal to Sat to nearest Data Center. Only 2 hops...

If you are close enough to the data center that one sat can see your terminal and the data center, I suspect that a terrestrial link will be faster.

[joek]

What Google and Starlink promised but yet to deliver. Terminal to Sat to nearest Data Center. Only 2 hops...

Sorry if a bit dense here: What did they promise that Kuiper might better fulfill? For example, the advantage vs. planting a ground station on or near an Internet Exchange (IXP)? A hop or two  isn't going to matter for the vast majority of the users and traffic.

And to the point I was responding to: How specifically would Kuiper benefit AWS? Not seeing it, maybe other than "Keeping up with the Jones's" (SpaceX et. al.). Which is reasonable; otherwise everyone might as well pack their bags and depend on Starlink, which don't think will happen.

[RedLineTrain]

What Google and Starlink promised but yet to deliver. Terminal to Sat to nearest Data Center. Only 2 hops...

Sorry if a bit dense here: What did they promise that Kuiper might better fulfill? For example, the advantage vs. planting a ground station on or near an Internet Exchange (IXP)? A hop or two  isn't going to matter for the vast majority of the users and traffic.

And to the point I was responding to: How specifically would Kuiper benefit AWS? Not seeing it, maybe other than "Keeping up with the Jones's" (SpaceX et. al.). Which is reasonable; otherwise everyone might as well pack their bags and depend on Starlink, which don't think will happen.

In reality, "packing their bags and depending on Starlink" is what is happening stealthily.  Starlink's ambitions and built reality continue to increase, while everyone else's is faltering.  Starlink is now touching the Ka-band, Ku-band, V-band, E-band, and soon W-band and Mid-band.  Satellite mass and dimensions continue to increase.  Starlink is putting its frequency licenses to use immediately and maximally, sometimes even building out capacity on spec with temporary licenses, like in the E-band.

[launchwatcher]

What Google and Starlink promised but yet to deliver. Terminal to Sat to nearest Data Center. Only 2 hops...

Sorry if a bit dense here: What did they promise that Kuiper might better fulfill? For example, the advantage vs. planting a ground station on or near an Internet Exchange (IXP)? A hop or two  isn't going to matter for the vast majority of the users and traffic.

A detailed discussion of where and how inter-provider peering is best done is likely not a great topic for this forum.   It's very technical but with complex economic/financial/political/competitive ramifications. 

Peering links need to be sized appropriately to avoid congestion.   My understanding is that a lot of peering (by traffic volume) happens 1:1 between big providers when they both have equipment and capacity in the same building, and I've seen reports that IXPs can get congested if the providers connecting don't use appropriate care in sizing their links. 

[joek]

A detailed discussion of where and how inter-provider peering is best done is likely not a great topic for this forum.   It's very technical but with complex economic/financial/political/competitive ramifications.

Agree.

Peering links need to be sized appropriately to avoid congestion.   My understanding is that a lot of peering (by traffic volume) happens 1:1 between big providers when they both have equipment and capacity in the same building, and I've seen reports that IXPs can get congested if the providers connecting don't use appropriate care in sizing their links.

Yes.

Now that we are done with that... [1] The operative question IMO is what do [LEO] satellite constellations bring to the table which makes them any different than any other carrier-provider? And more specifically, what advantage would Kuiper bring to AWS?


[1] Apologies if a bit short, but this is really basic stuff.

[DanClemmensen]

What Google and Starlink promised but yet to deliver. Terminal to Sat to nearest Data Center. Only 2 hops...

Sorry if a bit dense here: What did they promise that Kuiper might better fulfill? For example, the advantage vs. planting a ground station on or near an Internet Exchange (IXP)? A hop or two  isn't going to matter for the vast majority of the users and traffic.

A detailed discussion of where and how inter-provider peering is best done is likely not a great topic for this forum.   It's very technical but with complex economic/financial/political/competitive ramifications. 

Peering links need to be sized appropriately to avoid congestion.   My understanding is that a lot of peering (by traffic volume) happens 1:1 between big providers when they both have equipment and capacity in the same building, and I've seen reports that IXPs can get congested if the providers connecting don't use appropriate care in sizing their links.

Even worse: many (most?) ASs do intra-AS packet forwarding at layers below the network layer (e.g., a frame relay layer or an MPLS layer), and  and I strongly suspect there will be a specialized forwarding layer within each satellite constellation that has inter-satellite links. A mostly separate but related issue is the QoS and congestion management of the shared RF between the users in a spot and the satellite. Also take a look at modern cellular radio protocols (5G and earlier) that share the bandwidth in strange and exotic ways, all below the IP layer. All of this is out of scope for this thread.

[joek]

Even worse: many (most?) ASs do intra-AS packet forwarding at layers below the network layer (e.g., a frame relay layer or an MPLS layer), and  and I strongly suspect there will be a specialized forwarding layer within each satellite constellation that has inter-satellite links. A mostly separate but related issue is the QoS and congestion management of the shared RF between the users in a spot and the satellite. Also take a look at modern cellular radio protocols (5G and earlier) that share the bandwidth in strange and exotic ways, all below the IP layer. All of this is out of scope for this thread.

Yeah, but when they hit the Internet they're all the same. There may be sub-IP peering-transport-routing-whatever connections, but those are the walking dead. Been a few years since was involved with such, but even then it was clear: Do whatever you want with private peering-transport, but if you want to talk to us (the Internet), it's TCP-IP or go home (regardless of the underlying transport). Don't think that has changed?

[DanClemmensen]

Even worse: many (most?) ASs do intra-AS packet forwarding at layers below the network layer (e.g., a frame relay layer or an MPLS layer), and  and I strongly suspect there will be a specialized forwarding layer within each satellite constellation that has inter-satellite links. A mostly separate but related issue is the QoS and congestion management of the shared RF between the users in a spot and the satellite. Also take a look at modern cellular radio protocols (5G and earlier) that share the bandwidth in strange and exotic ways, all below the IP layer. All of this is out of scope for this thread.

Yeah, but when they hit the Internet they're all the same. There may be sub-IP peering-transport-routing-whatever connections, but those are the walking dead. Been a few years since was involved with such, but even then it was clear: Do whatever you want with private peering-transport, but if you want to talk to us (the Internet), it's TCP-IP or go home (regardless of the underlying transport). Don't think that has changed?

Of course. But within an AS, two routers that have a lower-layer transport may be on opposite sides of the world and layer-2 QoS may dominate. For satellite, user-to-satellite layer 2 will often dominate, and for 5G the 5G protocol dominates. My point is that thinking that all hops are IP is extremely oversimplified.

[launchwatcher]

The operative question IMO is what do [LEO] satellite constellations bring to the table which makes them any different than any other carrier-provider?

I can think of two things:

1) The ability to provision a modest amount of network capacity anywhere on the planet with no advance notice by delivering a small box.   You don't need to wait weeks or months or years for someone to pull a cable to where you need the bandwidth, or build a tower near where you need it.

2) the ability to provision a modest amount of network capacity to a moving vehicle.

You can get higher capacity (terabits/s) with better latency over fiber, but that takes longer, costs more, and requires a fixed delivery location.

And more specifically, what advantage would Kuiper bring to AWS?

That, I don't see.   I'd think it might well be more directly relevant to the stuff-delivery or consumer-IoT sides of Amazon than to the datacenter business.

[Asteroza]

The AWS value of Kuiper is a bit more multidimensional I think.


There is an outside chance that AWS is looking at capturing some vertical aspects, by trying to cut out some IP transport middlemen directly or indirectly via bullying. If kuiper gains D2D, then they could horsetrade D2D access to a carrier for priority transport of AWS hosted services (say Amazon Prime video/music). AWS has the Whispernet MVNO for their Kindles that could easily expand to eSIM data plans relatively easily, routed to AWS controlled terrestrial transport and backstopped with Kuiper D2D. Carriers could benefit by greatly expanding 5G high speed access via backstopping poorly connected/rural basestations via conventional Kuiper access through existing terminals (5G high speed is short range, requiring many more towers to be deployed, and that high speed access tower deployment capitalization is very much behind schedule outside of major urban centers).

[donaldp]

It appears that at the very least the Kuiper demonstration satalite thrusters are working well.

https://www.aboutamazon.com/news/innovation-at-amazon/amazon-project-kuiper-satellite-propulsion-system-november-2023-update

[FutureSpaceTourist]

https://twitter.com/thesheetztweetz/status/1725167092394004723

Amazon $AMZN announces full success of the Project Kuiper prototype internet satellites with no anomalies, and tests allowing for “streaming 4K videos” and video calls.

The company expects to begin launching commercial satellites in mid-2024.

https://www.cnbc.com/2023/11/16/amazon-kuiper-internet-satellites-fully-successful-in-testing.html

Amazon says its first Project Kuiper internet satellites were fully successful in testing
PUBLISHED THU, NOV 16 202310:00 AM EST

Michael Sheetz
@IN/MICHAELJSHEETZ
@THESHEETZTWEETZ

KEY POINTS

Amazon announced its pair of prototype internet satellites were fully successfully in operations and testing.

“Thirty days after launch, we’re streaming 4K videos, doing [video] calls, and shopping on Amazon.com,” Project Kuiper Vice President of Technology Rajeev Badyal told CNBC.

Amazon plans to begin building the first production Kuiper satellites in December and launch the first satellites for its network in mid-2024.

[FutureSpaceTourist]

https://twitter.com/spaceflorida/status/1729149018704576924

Explore @amazon @ProjectKuiper facility construction in recent drone shots at our Launch and Landing Facility. With a 2025 completion target, Amazon is making remarkable progress. The protoflight mission in October led to a 100% testing success, highlighting impressive strides!

[Twark_Main]

and shopping on Amazon.com

That's weird! I wonder why would they single out that one e-commerce site (out of thousands), specifically? 

And more important, why wouldn't they go to AliExpress and buy the exact same ████ for half the price? 

[FutureSpaceTourist]

Well …

https://twitter.com/breadfrom/status/1730670308678406468

OMG: Amazon's Project Kuiper secures a 3-launch deal with SpaceX

https://www.aboutamazon.com/news/innovation-at-amazon/amazon-project-kuiper-spacex-launch

how Search
NewsInnovation at Amazon
Amazon secures 3 launches with SpaceX to support Project Kuiper deployment
1 min
December 1, 2023
Written by Amazon Staff

Additional capacity will supplement existing launch contracts to support Project Kuiper’s satellite deployment schedule.

Amazon has signed a contract with SpaceX for three Falcon 9 launches to support deployment plans for Project Kuiper, Amazon’s low Earth orbit (LEO) satellite broadband network. Project Kuiper satellites were designed from the start to accommodate multiple launch providers and vehicles, allowing us to reduce schedule risk and move faster in our mission to connect unserved and underserved communities around the world. Our earlier procurement of 77 heavy-lift rockets from Arianespace, Blue Origin, and United Launch Alliance (ULA) provides enough capacity to launch the majority of our satellite constellation, and the additional launches with SpaceX offer even more capacity to support our deployment schedule.

SpaceX’s Falcon 9 is a reusable, two-stage launch vehicle designed for the reliable and safe transport of people and payloads into Earth orbit and beyond, and it has completed more than 270 successful launches to date. Project Kuiper has contracted three Falcon 9 launches, and these missions are targeted to lift off beginning in mid-2025.

Project Kuiper recently launched two prototype satellites, and tests from the mission have helped validate our satellite design and network architecture. We are preparing to start satellite manufacturing ahead of a full-scale deployment beginning in the first half of 2024, and we expect to have enough satellites deployed to begin early customer pilots in the second half of 2024.

To learn more about the Protoflight mission and next steps for the program, check out our latest mission updates.

[FutureSpaceTourist]

https://twitter.com/chenryspace/status/1730674003977580837

Didn't see this coming! I'm betting the recent news that Ariane 6 won't launch until mid 2024 (and may only launch twice next year) was probably making Kuiper's schedule hard to maintain. And with legacy vehicles in short supply, there was really only one place to go...

[DanClemmensen]

Didn't see this coming! I'm betting the recent news that Ariane 6 won't launch until mid 2024 (and may only launch twice next year) was probably making Kuiper's schedule hard to maintain. And with legacy vehicles in short supply, there was really only one place to go...

No surprise that they have finally acknowledged reality and will need to use Falcon 9. The math for this has been obvious for over a year.

There are two surprises, at least for me. First, they intend to start launching on Atlas V in H1 2024. I thought they planned to start in July 2024. Next, their first Falcon 9 launch is in 2025. That is still a success-oriented approach. They should start earlier to hedge against any further setbacks with their other launchers.  The only thing I can think of is that they are limited by satellite production and will not finish the eight Atlas V launches before then (about 280 satellites?) However if this is true it is hard to see how they can produce the 1618 satellites they need by July 2026.

Why just three Falcon 9? I speculate that their spreadsheets show a gap after the last Atlas V and the first availability of whatever they think is next. New Glenn? Vulcan?

[RedLineTrain]

Why just three Falcon 9? I speculate that their spreadsheets show a gap after the last Atlas V and the first availability of whatever they think is next. New Glenn? Vulcan?

My assumption is that there will be several additional three-launch follow on contracts.

[DanClemmensen]

Why just three Falcon 9? I speculate that their spreadsheets show a gap after the last Atlas V and the first availability of whatever they think is next. New Glenn? Vulcan?

My assumption is that there will be several additional three-launch follow on contracts.

Agreed. They will just keep hoping their other providers will magically keep their schedules.