Starlink : Markets and Marketing

[gongora]

That's just fearmongering from some telecom lobbyists.

[DigitalMan]

The reality is when Congress held a hearing regarding LEO satellite internet constellations, Starlink and OneWeb were represented, and rural service was one of the issues.

I don't see a reasonable argument to block them from participating based on my recollection of that hearing.

[OTV Booster]

You can see in Mark Handley's simulation, that even with the 400-satellite initial operating capability, a ground station in a covered area may see several satellites at once.

One figure of merit is coverage area divided by the number of satellites divided by the area each satellite covers.

With a 400 satellite constellation and 900km radius every spot on Earth is covered by an average of 2 satellites.  So every spot in the coverage area certainly is.  Since the initial constellations doesn't cover the whole earth the density of the coverage in the zones around 40 degrees north and south will be quite a bit higher.  I'd eyeball it at 10 or more.  A more accurate estimate could be made by subtract the uncovered area from the Earth's total surface area.

I don't know what oversubscription ratios are common for residential ISPs, though.  Somewhere between 2:1 and 10:1, I would guess, but that's a rather wide span.

The oversubscription ratio is an important number.  My guess is that it's higher than 10:1 and could be pushing 1000:1, at least at the level of a hub serving tens of thousands of residential customers.  Anybody have some actual data?

What are residential customers doing with GigaBit links?  Doing full disk backups multiple times a day?

I have a friend who’s a shut in for health reasons. He has Netflix high def running ~12-14 hours a day while he emails and texts to friends through a 100mb connection via small regional ISP. Then he bitches that they charge too much and sometimes throttle him. I want to throttle him too sometimes. He thinks he has a god given right to infinite bandwidth for free.

Bandwidth isn’t free. It has no mass, is colorless and odorless, and is essentially a magic thing to most people. It’s totally intangible so too many people think they should have as much as they want because it doesn’t seem to be a real thing to them. Until they don’t have it.

He gets almost homicidal when traffic cuts into the bandwidth and his TV gets pixelated and jerkey. Avoiding this is what people do with gigabit.

Phil

[niwax]

The oversubscription ratio is an important number.  My guess is that it's higher than 10:1 and could be pushing 1000:1, at least at the level of a hub serving tens of thousands of residential customers.  Anybody have some actual data?

What are residential customers doing with GigaBit links?  Doing full disk backups multiple times a day?

At the very high level, oversubscription ratios are massive. DE-CIX all-time peak was 100kbit/s per citizen, not to mention businesses, even though they have some 4x installed capacity. At the residential level, 10x is not unusual. When you look at the demographics and typical household makeup that makes a lot of sense. Going from 1000 8:00 PM Netflix junkies to an actual town of 10000 average households with 25200 people does not increase peak bandwidth by much.

Another fun tidbit, Cloudflare just sent out an email that the virus has caused overall data in their network to increase by about 10%, all the way up to 30% in Italy. The interesting thing here is that this is not a higher peak, but more load during the day.

[Ludus]

In terms of the Starlink Spinoff, Shotwell said something interesting:

https://mobile.twitter.com/joroulette/status/1237453859900948480?p=v :

"We still have a lot to do to see whether this is gonna work," Shotwell said. "It was just a way to potentially get employees... I'm not saying we're not gonna do it, but I'm just saying it is not in our thought process right now. It should not be news."

I read this as: getting employees that are not US citizens. Since SpaceX is building rockets, they have a high barrier to hire non US citizens. So a spinoff of Starlink could open the option to hire people from abroad easier.

My take is it’s about attracting employees to Starlink within SpaceX now. I think she’s saying Musk wants potential high value employees to be aware of the separate Starlink IPO possibility (adding value to stock options) but that they won’t devote any attention to it until Starlink is operational and making money.

[vsatman]

Assuming each user wants a speed of at least 20 Mbps, does that effectively mean Starlink is limited to serving 1000 customers per satellite at any one time?

Assuming 20 Mbit/s is the "advertised" bandwidth the customer buys, then no.  As an ISP, you always oversubscribe your capacity; you bet that only a portion of the users will be active, and using their full bandwidth, at any given time. 
I don't know what oversubscription ratios are common for residential ISPs, though.  Somewhere between 2:1 and 10:1, I would guess, but that's a rather wide span...  Likely depends on if there is effective competition for your customers as well; if customers don't have any practical alternative, the ISP can get away with much higher oversubscription and more dissatisfaction among its users.

My experience as satellite broadband internet provider - "The speed is nothing , traffic is all..."
According to https://www.telecompetitor.com/report-u-s-household-broadband-data-consumption-hit-268-7-gigabytes-in-2018/
"Average usage for households with flat-rate pricing was 282.1 GB per household,"  per month
282 Gb is  0,9 Mbit  for  24 hour 30 days usage..  or 100:1  oversubscription ratio for 100 Mbis connection .. (1 Mbit is 316 Gb  in month for 24 hour 30 days usage).

Starlink promise  17 Gbit  for 1 satellite   It mean  1 Sat can serve  17000 householdes...

But if all will see  Super Bowl  in 4K quality (4K online need 15 Mbit ) 1 sat can serve only 1000 householdes  (I assume 17 Gbit is 15 from Sat to terminal and 2 Gbit return channel to Sat)

 

[TomH]

Due to COVID-19, K-12 schools, colleges, and universities continue to cancel classes until fall, and that likely will be extended. Most are rushing to create online learning opportunities and classrooms. This will place a significant demand for bandwidth on the internet. Movie studios are moving current releases to streaming at higher prices than older releases, though this should not impinge on demand in the way online classes will.

Currently, the pandemic does not seem to have impaired Starlink production, Falcon production, or launch ability, but that could surely change in coming weeks and months. While the pandemic brings much of the economy to its knees, it concurrently brings a need for us to continue our jobs, learning, and other pursuits electronically from a distance. If Starlink can be quickly and successfully integrated into internet capacity, it could serve to ease the oncoming bandwidth demand. This transition is also slowed by the need for all students to have internet access, routers, computers, and browsers that are compatible with technology districts and colleges adopt. Another issue is whether a vaccine will cause this increased demand to crash, or whether a new paradigm will have been established. It is often said that danger and opportunity exist together within a crisis. It does seem to me that this is one of those kinds of situations. It will be interesting to see how this plays out for SpaceX and Starink.

[su27k]

Space should fare better against coronavirus than other industries, report says

Quilty Analytics, in its report, wrote that the COVID-19 crisis could drive demand for residential broadband — benefiting companies like EchoStar and Viasat — and stem cord-cutting that has winnowed subscriber numbers for traditional satellite television.

FCC Commissioner Geoffrey Starks advocated for a “connectivity stimulus” in a March 19 New York Times Op-Ed. Starks also called on U.S. broadband providers to keep Americans connected, including those needing low-cost options, while people telework and spend more time at home to limit their risk of catching and spreading the coronavirus.

[M.E.T.]

Assuming each user wants a speed of at least 20 Mbps, does that effectively mean Starlink is limited to serving 1000 customers per satellite at any one time?

Assuming 20 Mbit/s is the "advertised" bandwidth the customer buys, then no.  As an ISP, you always oversubscribe your capacity; you bet that only a portion of the users will be active, and using their full bandwidth, at any given time. 
I don't know what oversubscription ratios are common for residential ISPs, though.  Somewhere between 2:1 and 10:1, I would guess, but that's a rather wide span...  Likely depends on if there is effective competition for your customers as well; if customers don't have any practical alternative, the ISP can get away with much higher oversubscription and more dissatisfaction among its users.

My experience as satellite broadband internet provider - "The speed is nothing , traffic is all..."
According to https://www.telecompetitor.com/report-u-s-household-broadband-data-consumption-hit-268-7-gigabytes-in-2018/
"Average usage for households with flat-rate pricing was 282.1 GB per household,"  per month
282 Gb is  0,9 Mbit  for  24 hour 30 days usage..  or 100:1  oversubscription ratio for 100 Mbis connection .. (1 Mbit is 316 Gb  in month for 24 hour 30 days usage).

Starlink promise  17 Gbit  for 1 satellite   It mean  1 Sat can serve  17000 householdes...

But if all will see  Super Bowl  in 4K quality (4K online need 15 Mbit ) 1 sat can serve only 1000 householdes  (I assume 17 Gbit is 15 from Sat to terminal and 2 Gbit return channel to Sat)

As a non-engineer I would like to understand how current satellite internet providers can serve far larger markets from a single geostationary satellite? I don’t mean the larger geographic coverage, which is obvious, but the larger bandwidth presumably achieved per satellite than Starlink can manage? (Or is this assumption false?)

Why can Starlink not just make their satellites slightly bigger and increase bandwidth tenfold? Or is it a frequency/regulatory limitation rather than a hardware constraint?

Also, how does Oneweb’s bandwidth per satellite compare, considering they will be at a higher altitude, therefore covering a larger area and presumably more users per satellite?

[DigitalMan]

Assuming each user wants a speed of at least 20 Mbps, does that effectively mean Starlink is limited to serving 1000 customers per satellite at any one time?

Assuming 20 Mbit/s is the "advertised" bandwidth the customer buys, then no.  As an ISP, you always oversubscribe your capacity; you bet that only a portion of the users will be active, and using their full bandwidth, at any given time. 
I don't know what oversubscription ratios are common for residential ISPs, though.  Somewhere between 2:1 and 10:1, I would guess, but that's a rather wide span...  Likely depends on if there is effective competition for your customers as well; if customers don't have any practical alternative, the ISP can get away with much higher oversubscription and more dissatisfaction among its users.

My experience as satellite broadband internet provider - "The speed is nothing , traffic is all..."
According to https://www.telecompetitor.com/report-u-s-household-broadband-data-consumption-hit-268-7-gigabytes-in-2018/
"Average usage for households with flat-rate pricing was 282.1 GB per household,"  per month
282 Gb is  0,9 Mbit  for  24 hour 30 days usage..  or 100:1  oversubscription ratio for 100 Mbis connection .. (1 Mbit is 316 Gb  in month for 24 hour 30 days usage).

Starlink promise  17 Gbit  for 1 satellite   It mean  1 Sat can serve  17000 householdes...

But if all will see  Super Bowl  in 4K quality (4K online need 15 Mbit ) 1 sat can serve only 1000 householdes  (I assume 17 Gbit is 15 from Sat to terminal and 2 Gbit return channel to Sat)

As a non-engineer I would like to understand how current satellite internet providers can serve far larger markets from a single geostationary satellite? I don’t mean the larger geographic coverage, which is obvious, but the larger bandwidth presumably achieved per satellite than Starlink can manage? (Or is this assumption false?)

Why can Starlink not just make their satellites slightly bigger and increase bandwidth tenfold? Or is it a frequency/regulatory limitation rather than a hardware constraint?

Also, how does Oneweb’s bandwidth per satellite compare, considering they will be at a higher altitude, therefore covering a larger area and presumably more users per satellite?

I suppose its unfortunate the latency of GEO makes it unusable for anything but data replication across sites in any of the large software development teams I have worked with at various well-known companies. Remote access to resources would be too slow if you want to stay sane.

[niwax]

Assuming each user wants a speed of at least 20 Mbps, does that effectively mean Starlink is limited to serving 1000 customers per satellite at any one time?

Assuming 20 Mbit/s is the "advertised" bandwidth the customer buys, then no.  As an ISP, you always oversubscribe your capacity; you bet that only a portion of the users will be active, and using their full bandwidth, at any given time. 
I don't know what oversubscription ratios are common for residential ISPs, though.  Somewhere between 2:1 and 10:1, I would guess, but that's a rather wide span...  Likely depends on if there is effective competition for your customers as well; if customers don't have any practical alternative, the ISP can get away with much higher oversubscription and more dissatisfaction among its users.

My experience as satellite broadband internet provider - "The speed is nothing , traffic is all..."
According to https://www.telecompetitor.com/report-u-s-household-broadband-data-consumption-hit-268-7-gigabytes-in-2018/
"Average usage for households with flat-rate pricing was 282.1 GB per household,"  per month
282 Gb is  0,9 Mbit  for  24 hour 30 days usage..  or 100:1  oversubscription ratio for 100 Mbis connection .. (1 Mbit is 316 Gb  in month for 24 hour 30 days usage).

Starlink promise  17 Gbit  for 1 satellite   It mean  1 Sat can serve  17000 householdes...

But if all will see  Super Bowl  in 4K quality (4K online need 15 Mbit ) 1 sat can serve only 1000 householdes  (I assume 17 Gbit is 15 from Sat to terminal and 2 Gbit return channel to Sat)

As a non-engineer I would like to understand how current satellite internet providers can serve far larger markets from a single geostationary satellite? I don’t mean the larger geographic coverage, which is obvious, but the larger bandwidth presumably achieved per satellite than Starlink can manage? (Or is this assumption false?)

Why can Starlink not just make their satellites slightly bigger and increase bandwidth tenfold? Or is it a frequency/regulatory limitation rather than a hardware constraint?

Put simply, Starlinks weigh 200kg each, GTO sats are 5-7t. For now, Starlink is plenty big to serve their target market but can still be launched 60 at a time with the cheapest available rocket. There's no reason they wouldn't improve their sats with more capacity or further up the learning curve, although they might run into frequency issues first.

[gongora]

The difference in geographic coverage between the GEO and LEO sats is part of the reason for the difference in total bandwidth on a sat.  Since the GEO sats are covering a larger area, it's probably easier for them to reuse frequencies.  The larger size satellite gets them more power and more room for the electronics/antennas to create more spot beams.

There is a limited amount of spectrum available for the satellites to use.  Only certain frequencies are approved for satellite use.  As you get to higher frequencies there are greater affects from weather, so it will be interesting to see how V-band fares when it is introduced.  To get more bandwidth they create more beams of smaller size and reuse frequencies on non-adjacent beams.

[ZachF]

Assuming each user wants a speed of at least 20 Mbps, does that effectively mean Starlink is limited to serving 1000 customers per satellite at any one time?

Assuming 20 Mbit/s is the "advertised" bandwidth the customer buys, then no.  As an ISP, you always oversubscribe your capacity; you bet that only a portion of the users will be active, and using their full bandwidth, at any given time. 
I don't know what oversubscription ratios are common for residential ISPs, though.  Somewhere between 2:1 and 10:1, I would guess, but that's a rather wide span...  Likely depends on if there is effective competition for your customers as well; if customers don't have any practical alternative, the ISP can get away with much higher oversubscription and more dissatisfaction among its users.

My experience as satellite broadband internet provider - "The speed is nothing , traffic is all..."
According to https://www.telecompetitor.com/report-u-s-household-broadband-data-consumption-hit-268-7-gigabytes-in-2018/
"Average usage for households with flat-rate pricing was 282.1 GB per household,"  per month
282 Gb is  0,9 Mbit  for  24 hour 30 days usage..  or 100:1  oversubscription ratio for 100 Mbis connection .. (1 Mbit is 316 Gb  in month for 24 hour 30 days usage).

Starlink promise  17 Gbit  for 1 satellite   It mean  1 Sat can serve  17000 householdes...

But if all will see  Super Bowl  in 4K quality (4K online need 15 Mbit ) 1 sat can serve only 1000 householdes  (I assume 17 Gbit is 15 from Sat to terminal and 2 Gbit return channel to Sat)

As a non-engineer I would like to understand how current satellite internet providers can serve far larger markets from a single geostationary satellite? I don’t mean the larger geographic coverage, which is obvious, but the larger bandwidth presumably achieved per satellite than Starlink can manage? (Or is this assumption false?)

Why can Starlink not just make their satellites slightly bigger and increase bandwidth tenfold? Or is it a frequency/regulatory limitation rather than a hardware constraint?

Put simply, Starlinks weigh 200kg each, GTO sats are 5-7t. For now, Starlink is plenty big to serve their target market but can still be launched 60 at a time with the cheapest available rocket. There's no reason they wouldn't improve their sats with more capacity or further up the learning curve, although they might run into frequency issues first.

The empty mass of those GEO satellites are about 2-4 tonnes, Viasats will likely be towards the higher end.


The biggest reason is cost, Starlink satellites will literally cost about a thousand times less than one of Viasats stage 3 satellites, thanks to economies of scale.

[TomH]

As a retired teacher, I am currently hearing from former colleagues on the west coast as well as relatives on the east coast who are either teachers or work at universities, that school districts are in a race to put all students into online learning. Universities seem to have made the shift quickly, with students already expected to have online access as well as their own computers with compatible software. K-12 schools are not as far along, but seem to be scrambling to get Chromebooks into the hands of all students and access to the web. Many of my friends in various businesses have been instructed to work online from home. At home, I have noticed periodic bandwidth slowdowns, but bandwidth demand may soon begin to increase logarithmically along with the virus itself. As I said upthread, hopefully Starlink can help absorb some of that demand.

[Eka]

I wonder how much dark fiber is out there that can be lit up? When they lay fiber, they always lay more optical fibers than needed because it costs almost nothing to do it.

[Asteroza]

I wonder how much dark fiber is out there that can be lit up? When they lay fiber, they always lay more optical fibers than needed because it costs almost nothing to do it.

A lot of (longer distance) dark fiber is literally that. The lasers to light them up are physically not installed to cut capital costs. It's not something you can just switch on when you needed it. The physical plant costs are such that if you had lasers installed, it's usually simpler to keep it lit.

It looks like the Netflix video quality downshift is saving a good chunk of europe about 25% of Netflix sourced bandwidth, which is helping to ease the strain, but with classic ISP oversubscription of last mile resources, there's not much that can be done there. It has been helping according to major ISP's in the region. Netflix is considering deploying the bandwidth cut elsewhere on an as requested basis (read US). On such short notice though, local ISP's can't deploy Netflix OpenConnect CDN servers at their POP's to reduce their backhaul strain. The general video quality degradation step is their blunt solution for now.

[TGebs15]

As a retired teacher, I am currently hearing from former colleagues on the west coast as well as relatives on the east coast who are either teachers or work at universities, that school districts are in a race to put all students into online learning. Universities seem to have made the shift quickly, with students already expected to have online access as well as their own computers with compatible software. K-12 schools are not as far along, but seem to be scrambling to get Chromebooks into the hands of all students and access to the web. Many of my friends in various businesses have been instructed to work online from home. At home, I have noticed periodic bandwidth slowdowns, but bandwidth demand may soon begin to increase logarithmically along with the virus itself. As I said upthread, hopefully Starlink can help absorb some of that demand.

While this does seem like a valid use case for Starlink wouldn't this require one of two things:

1) They accelerate the production of user terminals and open up the network with limited capacity early.
2) The world will still be on lock down towards the end of the year when early service was expected to begin.

Elon seems reluctant to jump onto the panic ship but has been open to shifting plans in order to help with this event using his companies resources.

Edit: Decided to remove my personal opinion, didn't seem necessary for this discussion.

[indaco1]

Low latency if achieved could be paramount.

Trafic shaping could make LEO satellite networks competitive for backbone communications even when cost per byte is more, providing latency is less.

Low volume but important trafic, eg. control packets, requests, dns lookups, hadshaking etc. could be routed on the fast but expensive path.

Eg. suppose you are in Europe and request a web object in the US.  The request is few bytes, if you route it by satellite you could save, say, 30 milliseconds turnaround time, even if the content, that's big, is returned to Europe by fiber.

In other words, Starlink could improve performance of broadband communications if your ISP routes part of the trafic by satellite.

This could be an huge niche market.

[DistantTemple]

As a retired teacher, I am currently hearing from former colleagues on the west coast as well as relatives on the east coast who are either teachers or work at universities, that school districts are in a race to put all students into online learning. Universities seem to have made the shift quickly, with students already expected to have online access as well as their own computers with compatible software. K-12 schools are not as far along, but seem to be scrambling to get Chromebooks into the hands of all students and access to the web. Many of my friends in various businesses have been instructed to work online from home. At home, I have noticed periodic bandwidth slowdowns, but bandwidth demand may soon begin to increase logarithmically along with the virus itself. As I said upthread, hopefully Starlink can help absorb some of that demand.

While this does seem like a valid use case for Starlink wouldn't this require one of two things:

1) They accelerate the production of user terminals and open up the network with limited capacity early.
2) The world will still be on lock down towards the end of the year when early service was expected to begin.
Elon seems reluctant to jump onto the panic ship but has been open to shifting plans in order to help with this event using his companies resources.
Edit: Decided to remove my personal opinion, didn't seem necessary for this discussion.

The world hopefully will not be on lockdown in September.
IMO: However COVID-19 will not be over. A vaccine will not have been produced and widely given, natural immunity will not be widespread enough.
IMO: It might be something like this. There will at last be massive and cheaper testing. There will be extensive tracking of cases, and local hotspot lockdowns, of families, districts, certain schools and colleges etc. Even of cities. And definitely strict isolation of all suspected cases.

This will mean that online learning materials and platforms, as well as Zoom, and Skype, and online collaborative learning tools will need to be brought back in with hardly a days notice. Literally, "Class suspended, we will be online only for 14 days".

It will probably become a generally more widespread way of working. Universal (good) internet access will become essential.

Once students get used to it, it may even be more effective. Teaching classes of 30 is very Victorian, and prepares students for old fashioned factories, offices, and typing pools, not for taday's society! IMO (I'm a UK teacher)

[TomH]

I communicated today with former colleagues, relatives, and friends in OR, CA, and GA. All stated that teachers have been called in (physical presence required on campus) and told that they have to pivot, create online platforms, and adapt all curriculum for online learning. Some are in a district of 60,000+ students and have IT professionals and curriculum specialists to help do this. One is a new teacher with an emergency credential working in a tiny remote district in rural OR. She teaches a in school that goes from Kindergarten through 12th grade, all in one building, with about 100 students total. The district itself also is tiny. With no IT resources at all, they were simply told to figure it out and make it work.

High school students can be mature enough to learn independently. Primary students (typically K-3) are almost completely dependent on having a personal relationship with a mother-like figure who is right there. This is a new frontier in figuring out the psychology/sociology, but also for people who are not IT professionals, trying to become competent overnight. Teachers already had difficult jobs; this is going to be a Herculean task. I do hope ever diminishing bandwidth doesn't make everything all the worse.