Good Omens

[fjtorres]

Originally Posted by Quoth

With internet it's not scaleable to the mass market without servers at the ISP.
Also the percentage of internet used for ebooks is tiny. It can be over half for streaming TV.
The Cable company in many areas is the only viable broadband. Mobile, even 5G, can't replace fibre, dsl and cable for broadband and even less so for streaming video for a mass market. Basic physics.

The cable companies just won't be selling premium broadcast channels.

Netflix does (some) ISP servers with the bigger ISPs like Comcast but most other streamers don't.
It still works fine.

The Cablecos' internet monopoly isn't long for this world.

The satellite constellations can reach everyone everywhere and the new land-based wireless networks are "leveling up". They definitely can serve mass markets in cities with the milimeter wave segment of 5G.

Cable won't vanish but they'll be constrained from raising prices by competitors. The telcos are investing billions to go after that specific market.

It'll take a couple of years but STARLINK, for one will start signing up customers late this year. It's too bad the OneWeb went bankrupt over the crisis but Project Kuiper has Amazon money to work with so they'll hang around.

Edit: The Netflix edge servers don't exist because of capacity issues but because years ago Comcast discovered 60% of their evening taffic was Netflix and they tried to shake them down by threatening to throtle Netflix traffic so, rather than pay blackmail, Netflix installed the edge servers. Which they now use mostly to serve different national audiences. They help feed different content to different regions to protect against cyber attacks and accidental shutdowns but they're not mandatory. With modern streaming protocols bandwidth requirements are much lower.

Plus in the US there is still tons of excess backbone capacity in the form of "dark fiber" owned by independent companies who lease it out as required. During the current lockdown streaming demand is up by 40% with nary a glitch. And there's still plenty of headroom left.

Throttling these days is a no-no because net neutrality has become a major political issue.
And Comcast itself is a streamer now so throttling a competitor's traffic would be an invitation to the DOJ to take them to court.

Streaming is just getting started.

[Quoth]

"The satellite constellations can reach everyone everywhere and the new land-based wireless networks are "leveling up". They definitely can serve mass markets in cities with the milimeter wave segment of 5G."
Sorry but physically ONE street cabinet for VDSL, FTTH or Hybrid Fibre Cable can have more capacity than satellite for an entire region. About 10 people for a sector or 30 for a mast doing streaming (in fake HD) will max out a mast.
The 5G is largely hype. The higher faster bands are for stadiums or to replace cheaper WiFi in cafes and open plan offices. It's line of sight.
There are actual lies being peddled by some phone system vendors, some startup satellite outfits and the media.
The physics and mathematics mean that unless every 3rd lampost in a city is a mobile mast, mobile can't ever provide broadband. The contention is controlled by refusing or dropping connections.
There are broadly 3 kinds of satellite systems. They can sound fast, but that's a raw speed SHARED by 100,000 to a million users.

Cable, Fibre, DSL, VDSL are not going to be replaced for real broadband. That's fantasy. The satellite systems are competing with Mobile, typically mobile is better. Mobile is cellular so even 3G can do up to 10 times better in practice just by increasing the cell density. But unless the mobile is charged for on a data used + speed basis there is negative incentive to add more masts.

Throttling is complex. It absolutely has to happen on Mobile because otherwise there would be no voice calls. The Net neutrality debate in the USA is poisoned on both sides by competing corporate interests. The HFC (hybrid fibre cable) and Fibre (to home or cabinet) has so much capacity (about 1,000x of a phone mast and up to 100,000x per user of satellite) that it hardly needs caps or throttles on specific traffic.
Caps and throttles done purely for engineering reasons (not to favour your own product) are due to lack of capacity that on mobile would mean calls dropped or blocked. The 4G has no efficient native voice at all, so actually has poorer capacity for voice calls than 2G or 3G per Mega Hertz, the operator calls use VOIP, so there is no reason to limit VOIP on 4G.
Traditional cable and DSL is more complex and it's true that some operators throttled traffic for competitive reasons rather than engineering. The simple solution is for ISPs (Internet Providers) to be forbidden to sell any other service on their platform.
Also more fibre.
Satellite is for people in the wastelands, jungle, deserts. Not even Africa now as they have mobile and fibre.
Mobile should not be used by people sitting at home or in an office. That means less speed or no connection for people actually MOBILE. All homes and offices should have fibre. Apart from 100 to 1000x faster, it's always on and uses about 20x less electricity for 1000x speed! Decent masts need fibre feeds anyway.

The growth in streaming can only be supported by modern cable (HFC), VDSL or Fibre. Satellite and Mobile can only support a handful of people per mast sector (typically max 3 sectors) or per satellite per town/village as the satellite covers a large area. The very low orbit ones only solve latency, they are poor capacity per user if economically subscribed and each one is only available for maybe 10 minutes, that's why they are needing a big fleet. Being low they have a short operational life compared to the Geostationary satellites (which have horrible latency).

[pwalker8]

Originally Posted by Quoth

"The satellite constellations can reach everyone everywhere and the new land-based wireless networks are "leveling up". They definitely can serve mass markets in cities with the milimeter wave segment of 5G."
Sorry but physically ONE street cabinet for VDSL, FTTH or Hybrid Fibre Cable can have more capacity than satellite for an entire region. About 10 people for a sector or 30 for a mast doing streaming (in fake HD) will max out a mast.
The 5G is largely hype. The higher faster bands are for stadiums or to replace cheaper WiFi in cafes and open plan offices. It's line of sight.
There are actual lies being peddled by some phone system vendors, some startup satellite outfits and the media.
The physics and mathematics mean that unless every 3rd lampost in a city is a mobile mast, mobile can't ever provide broadband. The contention is controlled by refusing or dropping connections.
There are broadly 3 kinds of satellite systems. They can sound fast, but that's a raw speed SHARED by 100,000 to a million users.

Cable, Fibre, DSL, VDSL are not going to be replaced for real broadband. That's fantasy. The satellite systems are competing with Mobile, typically mobile is better. Mobile is cellular so even 3G can do up to 10 times better in practice just by increasing the cell density. But unless the mobile is charged for on a data used + speed basis there is negative incentive to add more masts.

Throttling is complex. It absolutely has to happen on Mobile because otherwise there would be no voice calls. The Net neutrality debate in the USA is poisoned on both sides by competing corporate interests. The HFC (hybrid fibre cable) and Fibre (to home or cabinet) has so much capacity (about 1,000x of a phone mast and up to 100,000x per user of satellite) that it hardly needs caps or throttles on specific traffic.
Caps and throttles done purely for engineering reasons (not to favour your own product) are due to lack of capacity that on mobile would mean calls dropped or blocked. The 4G has no efficient native voice at all, so actually has poorer capacity for voice calls than 2G or 3G per Mega Hertz, the operator calls use VOIP, so there is no reason to limit VOIP on 4G.
Traditional cable and DSL is more complex and it's true that some operators throttled traffic for competitive reasons rather than engineering. The simple solution is for ISPs (Internet Providers) to be forbidden to sell any other service on their platform.
Also more fibre.
Satellite is for people in the wastelands, jungle, deserts. Not even Africa now as they have mobile and fibre.
Mobile should not be used by people sitting at home or in an office. That means less speed or no connection for people actually MOBILE. All homes and offices should have fibre. Apart from 100 to 1000x faster, it's always on and uses about 20x less electricity for 1000x speed! Decent masts need fibre feeds anyway.

The growth in streaming can only be supported by modern cable (HFC), VDSL or Fibre. Satellite and Mobile can only support a handful of people per mast sector (typically max 3 sectors) or per satellite per town/village as the satellite covers a large area. The very low orbit ones only solve latency, they are poor capacity per user if economically subscribed and each one is only available for maybe 10 minutes, that's why they are needing a big fleet. Being low they have a short operational life compared to the Geostationary satellites (which have horrible latency).

I agree with you. You can't beat wired for bandwidth. Satellite is for those who aren't close enough to make running wire economical. 5G is just faster LTE. Good enough if you are streaming on your phone, but not really designed to be a replacement for wired with a family of four all streaming to different TV's at 4K. They are different solutions for different problems. Inside the business, 5G has long been talked about mostly as device to device protocol. Some talk about using it to get from the hub to the house rather than send someone out to run wire.

[Quoth]

The 5G air interface will make a difference only for stadiums, racecourse and similar. The main and more useful aspects of it are easier to manage infrastructure. It's of no advantage for ordinary mobile compared to 4G. Even 4G in a 5MHz channel is about the same as 3G if both have 5 users. Each user would get about 1/5th of the claimed max speed if signal was perfect. The 3G uses inefficient CDMA, which was first used for secure point to point military encryption. The USA 2G used that, typically a 1MHz channel, but Europe 2G used the superior GSM, but only 200 kHz channels. Qualcomm and others with CDMA patents got the obsolete 3G agreed. The problem with CDMA is that as more users are added the cell shrinks and some are dropped. With 20 users, the GSM or 4G users get about 1/20th each if all else is equal. But with CDMA-1 / EVDO (USA 2G) and 3G, with 20 users the cell is about 1/4 the area and speed per user could be 1/40th. So that is why 4G was brought in. Also 3G ONLY has 5MHz channels, the 4G can do 5, 10 or 20 MHz. Also 4G can now use two channels from different masts. The quoted maximum speed is of course the totally rare 16x peak 3G speed, assuming NO other users on TWO mast sectors and a perfect signal.
Speed on EVERY WIRELESS system drops with distance. Double distance = 1/4 speed. Shannon - Nyquist Law, proposed as a theory in 1948. It's mathematics. It can't be beaten.
So the ONLY way 5G can go faster and for more people (at best, with a perfect protocol EVERY wireless system is shared, so 10 users get a piece of speed each!) is a higher band with wider channels. BUT PHYSICS! The BEST bands are 900MHz and 1800MHz (1.8 GHz). The 2.1 GHz is much less reach and more easily blocked. Compare 2.4GHz and 5.8 GHz WiFi. The higher WiFi band is almost useless outside the room and is best for an open plan office. The 2.4 GHz WiFi will work to the garden workshop / shed, if there are facing windows (airpoint in house & phone in shed).
The 800 MHz is poor because the cell size is too big. The 700 MHz should have stayed with TV for 4G. It's huge uncontrollable cells and only good if you are only one of a couple of users.
The 5G infrastructure and a real 5 G phone allows calls and data to move between bands and even WiFi, GSM and 3G without dropping the call or connection. In theory. The big channels are on higher bands than are any use for mobile. Great for a stadium or racecourse.

Some talk about using it to get from the hub to the house rather than send someone out to run wire.

Pointless for delivery to home in rural, we've had brilliant dedicated point to point as good as fibre with dishes for years and point to many cheaply using rooftop aerials at speeds between ADSL2+ and VDSL, real broadband, for over 15 years!
Anywhere that has wired Electricity can have fibre. It's cheaper now to install fibre than copper phone cable and less likely to be stolen. Almost no scrap value.

So 5G only helps stadiums, racecourses and Mobile Telcos PR. The real infrastructure features save them a little money and make user experience more reliable. Any office or cafe replacing Fibre/cable/DSL fed WiFi by mobile Operator 5G is going to have the same or worse performance and a much bigger bill, or the cafe people wil be using up their allowance.
Even unlimited mobile is not. There is no such thing. There HAS to be a fair use clause to clobber the heavy users because then other people are sometimes not even getting connected and might change operator. Want the best mobile? Check they have masts or a mast sharing agreement in your area. Then pick from ones that offer lower caps, or no unlimited. Maybe even ones that DON'T offer a top of the range phone free. The customers are paying on the data charges. Or the operator is skimping on mast rollout. There are no free phones.
This was apparently produced in 2009. The predictions about 4G turned out to be true:
http://www.radioway.info/comparewireless/

[fjtorres]

Originally Posted by Quoth

Anywhere that has wired Electricity can have fibre. It's cheaper now to install fibre than copper phone cable and less likely to be stolen. Almost no scrap value.

That will vary by country.
There is "rural", rural, and "RURAL!!!" .

In the US there are many places with less than 10 customers per mile and often dozens of miles between customers. At US$20,000 per mile for the turnk and $600 to connect the customer, the result is extremely high prices and extremely low speeds.

The need is big enough that LEO satellite constellations are projected to rake on $30Billion a year.

https://www.youtube.com/watchv=tufs0zowybg

As for 5G there is more to it than just "slightly better LTE". It has multiple layers, some low bandwidth/low latency (for autonomous auto communicstions and IoD) others very high bandwith/short range, using milimeter wave transmitters every few hundred feet. Some of the wireless carriers are instaling their nodes on light posts and traffic signals in urban and suburban areas and wiring up individual developments.

Both Verizon and AT&T are launching 5G Home Broadband in the US:

https://www.verizon.com/5g/home/

https://www.howtogeek.com/428337/not...and-explained/
The cable companies may think they have an eternal monopoly to mint money but the wireless and big tech companies intend to dispute this. That monopoly money is a big inducement. Starlink, in particular, is going after the premium, ultra low latency customers (bog bucks), and opportunistically picking off regular bandwidth customers elsewhere. The system needs massive constellations to serve the big bucks customers and in the process provide excess capacity they can sell cheap. (That is how Amazon ramped up AWS, by renting out their own excess capacity. It'll work for Starlink and probably Kuiper, too.)

As usual, big tech is playing "what's mine is mine, what's yours is up for grabs".

[jbjb]

Originally Posted by Quoth

Speed on EVERY WIRELESS system drops with distance. Double distance = 1/4 speed. Shannon - Nyquist Law, proposed as a theory in 1948. It's mathematics. It can't be beaten.

That's not correct, I'm afraid.

The Nyquist-Shannon theorem concerns the relationship between sample rate and signal bandwidth when sampling continuous-time signals. It's got nothing to do with distance.

Essentially, it says that in order to be able to accurately reconstruct a sampled continuous-time signal you need to sample it at greater than twice the rate of the highest frequency component of that signal.

[pwalker8]

Originally Posted by Quoth

The 5G air interface will make a difference only for stadiums, racecourse and similar. The main and more useful aspects of it are easier to manage infrastructure. It's of no advantage for ordinary mobile compared to 4G. Even 4G in a 5MHz channel is about the same as 3G if both have 5 users. Each user would get about 1/5th of the claimed max speed if signal was perfect. The 3G uses inefficient CDMA, which was first used for secure point to point military encryption. The USA 2G used that, typically a 1MHz channel, but Europe 2G used the superior GSM, but only 200 kHz channels. Qualcomm and others with CDMA patents got the obsolete 3G agreed. The problem with CDMA is that as more users are added the cell shrinks and some are dropped. With 20 users, the GSM or 4G users get about 1/20th each if all else is equal. But with CDMA-1 / EVDO (USA 2G) and 3G, with 20 users the cell is about 1/4 the area and speed per user could be 1/40th. So that is why 4G was brought in. Also 3G ONLY has 5MHz channels, the 4G can do 5, 10 or 20 MHz. Also 4G can now use two channels from different masts. The quoted maximum speed is of course the totally rare 16x peak 3G speed, assuming NO other users on TWO mast sectors and a perfect signal.
Speed on EVERY WIRELESS system drops with distance. Double distance = 1/4 speed. Shannon - Nyquist Law, proposed as a theory in 1948. It's mathematics. It can't be beaten.
So the ONLY way 5G can go faster and for more people (at best, with a perfect protocol EVERY wireless system is shared, so 10 users get a piece of speed each!) is a higher band with wider channels. BUT PHYSICS! The BEST bands are 900MHz and 1800MHz (1.8 GHz). The 2.1 GHz is much less reach and more easily blocked. Compare 2.4GHz and 5.8 GHz WiFi. The higher WiFi band is almost useless outside the room and is best for an open plan office. The 2.4 GHz WiFi will work to the garden workshop / shed, if there are facing windows (airpoint in house & phone in shed).
The 800 MHz is poor because the cell size is too big. The 700 MHz should have stayed with TV for 4G. It's huge uncontrollable cells and only good if you are only one of a couple of users.
The 5G infrastructure and a real 5 G phone allows calls and data to move between bands and even WiFi, GSM and 3G without dropping the call or connection. In theory. The big channels are on higher bands than are any use for mobile. Great for a stadium or racecourse.

Pointless for delivery to home in rural, we've had brilliant dedicated point to point as good as fibre with dishes for years and point to many cheaply using rooftop aerials at speeds between ADSL2+ and VDSL, real broadband, for over 15 years!
Anywhere that has wired Electricity can have fibre. It's cheaper now to install fibre than copper phone cable and less likely to be stolen. Almost no scrap value.

So 5G only helps stadiums, racecourses and Mobile Telcos PR. The real infrastructure features save them a little money and make user experience more reliable. Any office or cafe replacing Fibre/cable/DSL fed WiFi by mobile Operator 5G is going to have the same or worse performance and a much bigger bill, or the cafe people wil be using up their allowance.
Even unlimited mobile is not. There is no such thing. There HAS to be a fair use clause to clobber the heavy users because then other people are sometimes not even getting connected and might change operator. Want the best mobile? Check they have masts or a mast sharing agreement in your area. Then pick from ones that offer lower caps, or no unlimited. Maybe even ones that DON'T offer a top of the range phone free. The customers are paying on the data charges. Or the operator is skimping on mast rollout. There are no free phones.
This was apparently produced in 2009. The predictions about 4G turned out to be true:
http://www.radioway.info/comparewireless/

5G was never about rural. Rural is about distance, and 5G is much shorter ranged than LTE. The last mile that I was talking about was purely in neighborhoods with enough density to justify putting in a microcell per block. My sister who lives out in the boonies, finally got a T1 line put in (yea, I know the pedantic insist that it can't be a T1, but that's what AT&T calls it, so that's what I call it) to replace their satellite service. They could get up to 100 Mbps, but they don't particular need that level speed.

Streaming video changes the economics of wireless. The assumption with wireless is that everyone won't be using it at the same time. That's why wireless throttles people after a certain amount of usage during busy times.

[Quoth]

Originally Posted by pwalker8

The last mile that I was talking about was purely in neighborhoods with enough density to justify putting in a microcell per block.
. . .
Streaming video changes the economics of wireless. The assumption with wireless is that everyone won't be using it at the same time. That's why wireless throttles people after a certain amount of usage during busy times.

Except the solutions to that are cheaper than 5G, can be faster and have existed for 15+ years!

Mobile was always only meant to be for MOBILE use, short bursts of looking up a web page. Or short video calls. It was never meant as an alternative to wired broadband of ANY kind. The contention is terrible and unlike fixed wireless solutions the operator can only control contention by dropping connections, blocking connections and throttling.
Except the former incumbent in Ireland gutted DSL by selling mobile at a monthly rate lower than the line rental (before you even added a broadband sub), to match Three, who for years never made profit but sold cheap to build customer base. Several failures there. So line rental fell from 82% to close to 50%. It was a gift to UPC cable when they took over Chorus & NTL (now renting Virgin name, a subsidiary of Liberty Global).

There are other problems with streaming too. It was meant to offer "long tail", instead to attract customers it's competing with new releases on other media.

[Quoth]

You are mixing up two things. The law on information is often simply called Shannon's Limit. The Sample rate issue is usually called the Nyquist Limit.

Originally Posted by jbjb

That's not correct, I'm afraid.
The Nyquist-Shannon theorem concerns the relationship between sample rate and signal bandwidth when sampling continuous-time signals. It's got nothing to do with distance.

I see they are now calling it Shannon-Hartley. Quite obviously from the context that's what I was referencing.
Go read it. It covers EVERY aspect of the signal channel. With more distance there is less signal. The Inverse Square law.
With less signal the signal to noise is worse, so mathematically you have to have less bits per symbol.
You can't easily have a bigger aerial or more power in the handset, so roughly you get a 1/4 speed when you double the distance.

The Nyquist frequency limit is what you may be thinking of. You have to sample at twice the highest frequency. The I & Q quadrature systems only seem to have a sample rate = max f, but because the samples are 90 degrees offset and there are twice as many samples, it's effectively twice the rate.

The names of these things vary:
SAMPLING https://en.wikipedia.org/wiki/Nyquis...mpling_theorem

Information Transmission https://en.wikipedia.org/wiki/Shanno...artley_theorem

Shannon Limit https://en.wikipedia.org/wiki/Noisy-...coding_theorem

So I'd rather watch Good Omens on HD Satellite than via streaming. But I might buy the disc when it's released.

[jbjb]

Originally Posted by Quoth

You are mixing up two things. The law on information is often simply called Shannon's Limit. The Sample rate issue is usually called the Nyquist Limit.

No, I'm not mixing up anything. The Nyquist-Shannon theorem refers to sampling rate and bandwidth:

(See https://en.wikipedia.org/wiki/Nyquis...mpling_theorem)

I see they are now calling it Shannon-Hartley. Quite obviously from the context that's what I was referencing.

Shannon-Hartley is indeed more relevant, but again even that doesn't mention distance. Shannon-Hartley theorem defines channel capacity in terms of signal to noise ratio. The SNR may indeed fall off with the square of distance in most cases, but when you make claims about "EVERY WIRELESS system" (your capitalisation) you need to be a bit more careful. In the presence of non-isotropic transmitters, beam-forming, collimated beams, focused beams etc. etc. that's not necessarily the case. Even in the cases which are inverse-square, the effective origin of the beam isn't necessarily physically coincident with the transmitter, so in terms of distance from the transmitter the performance can be much better than inverse square.

It covers EVERY aspect of the signal channel.

You're being a bit over-free with the "EVERY" - Shannon-Hartley doesn't cover everything.

With more distance there is less signal. The Inverse Square law.

Not necessarily inverse-square, as discussed above.

[Apache]

Talk about a hijacked thread.
Apache

[fjtorres]

Originally Posted by Apache

Talk about a hijacked thread.
Apache

And nobody answered whether there's a sequel coming to Prime Streaming.
(Apparently not.)