Starlink Beta Has Outperformed Most Internet in the US | Hacker News Reader

Starlink Beta Has Outperformed Most Internet in the US

futurism.com

79 pointskarimford90 comments

What blows my mind is how a lot of parts in highly developed countries like the US or Germany don't have access to a broadband internet connection.

We had a lot of investment from the EU to install optical fibre everywhere, even in small towns. Now we have even small companies that only operate locally with very, very competitive prices.

At least the US has the excuse of being huge, which does make it much harder to provide good coverage. We don't have that excuse here in Germany.

The situation isn't entirely terrible in Germany, but extremely variable depending on your location. I have 1Gbit for 40 EUR per month, but there are certainly places that aren't even remote that get no reasonable broadband.

Yeah, the US is huge, but technologically we could say they are the most advanced country in the world. All the big boys are there. I would expect a lot more from them.

And in some places, even big cities, people don't even have a choice as there's only company.

> At least the US has the excuse of being huge, which does make it much harder to provide good coverage.

Why does that excuse America? The continent of Europe is also huge. It isn’t like this stuff is installed by one solitary overworked technician going house-to-house, big countries (and regions) can employ more people to do more work, and can raise more money from more customers.

The US has around 15% of the population density of Germany. It's much easier to provide broadband to larger cities than to rural areas.

I guess it might be also pretty hard if it's city centre full of old buildings.

Norway has overall low population density, and seems to do fine.

148,729 mi² vs 3.797 million mi²

15 people per square kilometer (Norway) vs 34 people per square kilometer (continental us).

With corresponding higher amount of people to do the cabling. (Which is still irrelevant, nobody complains about the lack of gigabit ISPs in the Rockies or the Wyoming countryside, they complain that you can't get it in coastal or midwestern suburbia where the vast majority of Americans live.

They have a ridiculous amount of oil for their size.

That being said, they've also done a pretty good job in managing that money (which is rare for natural resource rich countries).

The population spread of Norway more closely resembles that of Canada than that of the US, though. It's very difficult to inhabit large swathes of land in the north and center of Norway, so most people live in the southern and western parts of the country near the coast.

In the US, on the other hand, remote places are a lot more inhabitable. There are just more remote towns in the US, with more people owning more land and being spread out more as a result.

Of course, this doesn't excuse the fact that some dense towns in the US a few miles from a major city still have to deal with shitty broadband because no ISP is willing to invest in hooking them up to FttH. On average, though, the US will probably do worse than Norway for the foreseeable future because of the difference in where and how people live.

Okay, and how's the internet out there, in the more rural parts of Norway?

But yeah the bigger issue is that often even in urban or suburban areas, the internet is still bad.

That’s disingenuous. You don’t need broadband in the middle of the desert where there is no population. You need it in cities and towns where the density is much more similar to Europe etc.

That's true, but the USA is famous for huge, sprawling suburban rings around its cities (vs Europe where the impression is far more 'inner city apartments').

Sure, but US has higher urbanisation rate. There's just no excuse not to link up your urban areas.

It's even worse if you're a company in Germany. The pricing for business lines (and the speeds you get) are atrocious.

Here in Berlin, there's not even fiber for consumers yet (there probably is for commercial). Meanwhile, in my native Zurich there's fiber everywhere and you get 10x the speed for the same price.

> At least the US has the excuse of being huge, which does make it much harder to provide good coverage.

That could be an excuse if you ignore the fact that the terrible service is not limited to rural regions in the US. It's often really expensive and slow even in places that are packed with people.

It's an excuse all the same; if you can put electricity and phone lines on every address, fiber isn't that much harder. It' just that major ISPs in the US are sociopaths and don't like the margins you'd get with such a setup. Slow, overbooked connection hubs and for-profit-DPI and data collection (and sales) on an ISP level is the norm. Even net neutrality isn't something the government is capable of accepting and enforcing.

What part of the EU are you in?

Spain

Yeah, Spain is interesting. Or at least the way it's done in Barcelona. I don't know who owns the fiber (the city?), but it seems that fiber was run to every building/apartment, and when you sign up for service with one of the ISPs, they just give you a combo ONT+router and presumably plug the other end of the fiber line into their OLT. I'm guessing all the ISPs have their OLTs together so they can swap the lines appropriately.

That being said, it seems the networks have a lot of congestion at the neighborhood level (equivalent to oversubscribed DOCSIS nodes), at least in my experience in several neighborhoods in Barcelona. This became much more visible when COVID hit and Netflix and such had to limit quality in Europe as a result.

I only have experience with Movistar and Vodafone fiber, I Haven't used Orange's so I'm not sure if they're any better.

Don't they use PON there? Seems expensive to have to run an individual fiber line to every household, rather than sharing a single fiber (with branches).

It's hard to not see fibre internet as a waste of money right now given 4G/5G costs coming down dramatically.

For me personally, 4G is extremely fast and cheap while being, importantly, low-latency.

Even gaming and other latency sensitive applications are possible on 4G/5G. Why not just roll out 5G networks?

Eventually, 5G will outperform fibre on a cost and speed basis (for a long time my FTTH home internet was slower than my 4G connection by a large margin). Of course, I mean practical terms rather than theoretical. Practically, fibre speeds are usually limited by 1Gbps-maximum middleboxes in community fibre cabinets that are not cheap to upgrade.

I disagree. 4G is severly limited in capacity so speed will vary everywhere, and same will apply to 5G too. Fiber has incredible capacity and can be upgraded by changing OLT and ONT equipment on both sides. It can also run multiple protocols.

2 Gbps XGPON2 is starting to roll out in my country and it's a rather simple upgrade (but costly). If your ISP isn't offering gigabit on fiber, it's purely motivated by greed and lack of competition.

Also I'm not aware of any "community fiber cabinets" limited to 1 Gbps. Are you talking about GPON or Ethernet networks? Ethernet is not FTTH but FTTN, like DOCSIS or VDSL can be.

I have GPON FTTH, as in I can see the fiber strand coming into my apartment. If I call providers and offer to buy beyond 1Gbps for ANY PRICE, even on enterprise plans, they can't do it.

The network is not technically capable of it.

It doesn't necessarily mean the network is not technically capable, they probably just don't have a plan with faster speeds.

GPON is 2.4/1.2 Gbps and it's shared with multiple users so it's incredibly unlikely that any part if the core infrastructure is limited to Gigabit. Only the CPE is.

You can deploy 2 Gbps service on GPON though the quality will be low-ish, or you can run multiple versions of xPON standards on a single fiber network and offer > 1 Gbps to users who you switch to XGPON2. Others stay with existing CPE and plans and GPON as if nothing changed.

I've had 5G home boradband for roughly the last 6 months. Generally getting speeds of 300+ Mbps which is better than the 20 Mbps I can get at my property via copper cables (can't get fibre installed despite being in a large city).

However, whilst the speed is great, the reliability is not. Its currently been disconnected for the last 48 hours with no timeframe given for repairs to the mast. So I'm having to tether of 4G on my phone (on another network).

tl;dr

5G speeds ++ 5G uptime --

For cellular networks the general tradeoff is speed vs range - the higher speed protocols have much shorter range, so you need a more dense network of cell towers. Noone is going to splatter the whole countryside with 4G LTE high-speed coverage, this is going to cover the urban areas only. It's simply not practical to do it for low density areas, you'd have to put up a whole cell tower to cover just a couple people and they aren't going to pay for that; you'd use frequencies and protocols that get longer range but don't get as much speed.

It's worth noting that 5G explicitly combines very different techniques with very different properties (e.g. high-bandwidth short range systems and low power slow networks for things like smart meters), and the 5G systems that are competitive with fibre are essentially designed for dense metropolitan areas, USA won't cover all the suburbs with high-speed 5G.

> Noone is going to splatter the whole countryside with 4G LTE high-speed coverage

You are welcome to come check out our excellent unlimited and uncapped 4G coverage in rural Finland.

Proper network coverage isn’t an issues in countries with functioning telecom regulation and coverage minimums in frequency licenses.

Unlimited & uncapped, yes.

Excellent, doesn't match my experience. I'm not sure if I've been too rural or not rural enough.

4G is fast and cheap when people's bulk data is being offloaded onto wired networks due to data caps. If you put literally all consumer usage on 4G/5G you're going to have to raise the density dramatically, and then you're pulling fiber to every neighborhood anyway, might as well pull them straight into the homes when you're at it and then you're future-proof to 10 Gbit+

I live in rural Utah. Population of our entire county (45 minute radius), is 50k.

For $60/month I get 1 GB down, and about 100MB UP. I have google fi, and yesterday I could barely get reddit gifs to load (it was slow for some reason)... usually it's better, but my FIBER IS WAY BETTER.

There is active work and lots of money spent by large ISPs and telecoms to prevent deployment of additional networks, and to make it as difficult and expensive as possible for new entrants to start up.

It really should be a national shame, how much the US has permitted AT&T and Verizon and Comcast and Time Warner et al to derail the correct functioning of a market so insanely critical to the growth of prosperity (both personal as well as the wider economy).

This is what happens when you let the state decide who is allowed to offer certain types of services: pretty soon the large, well-funded incumbents take over the state-run process about deciding who is allowed in the market, and enact policies and regulations that make it cost millions to serve even a single customer. This is peanuts for large existing companies with billions of dollars of recurring revenue, but impossible for new entrants.

The guy who runs the FCC and decides this kind of stuff, for example, used to work for Verizon, one of the largest telcos in the country.

In most of the US, it is basically illegal to start a new ISP.

Don't feel too bad, in Australia one Government party actively sabotaged the national optic fiber infrastructure project of the other Government party.

Australia's current government has literally held the country back a decade in network infrastructure progress, one of the key foundations of any country's economic future.

It IS a national shame.

A lot of municipalities in the US have high pole attachment fees, or even worse, sell a monopoly on pole attachment to telecoms.

13 states cap pole attachment fees at cost or at a fixed number. [1] I'd be interested to see this pass in Pennsylvania.

[1] https://www.fcc.gov/sites/default/files/ad-hoc-commitee-surv...]

> There is active work and lots of money spent by large ISPs and telecoms to prevent deployment of additional networks

You mean lobbying by telcos to make permitting and other things expensive in municipalities and counties.

> This is what happens when you let the state decide who is allowed to offer certain types of services

Yes. Absolutely this.

"[...]a lot of parts in highly developed countries like [...] Germany don't have access to a broadband internet connection."

Huh? Where do you get that from? "A lot of parts", even? I've lived in one of the most rural areas with basically only villages and small towns (<10000 pop) around. When I lived in my village of 600, we got broadband access in 2005.

You might have different definitions of broadband.

That's not how it works though. The International Telecommunication Union's definition is relevant here. For instance, in 2010, it was: "[...]downstream speeds equal to, or greater than, 256 kbit/s...".[1]

There was also talk about increasing it to 2 Mbit/s some years ago.[2]

Not that that change would make a difference if you look at broadband accessibility for Germany here.[3]

Point is: What was written was a load of bollocks. There were no facts to back up the statement but people lapped it up anyway.

[1] https://www.itu.int/ITU-D/ict/material/TelecomICT_Indicators...

[2] https://www.itu.int/en/ITU-T/committees/scv/Documents/T17-SC...

[3] https://www.bmvi.de/DE/Themen/Digitales/Breitbandausbau/Brei...

> highly developed countries like the US or Germany

In Germany this is due to regulation. Deutsche Telekom has no incentive to lay cables, as they will have to share the infrastructure with competitors at a price that makes doing so unattractive.

Sweden has the same regulation but we got widespread fiber anyway

> the same regulation

As in, you compared the regulated prices?

It's been a while since I lived in Sweden but I didn't know anyone with fiber from the incumbent, there were lots and lots of national and local competitors (power companies, municipal networks, other telcos, internet startups, local co-ops)

>Deutsche Telekom has no incentive to lay cables, as they will have to share the infrastructure with competitors at a price that makes doing so unattractive.

Seems like a pretty severe failure of government if they actively disincentivised investment into fiber tech...

That is what governments do, make infrastructure more expensive to build.

No, it's due to privatization. The Bundespost would simply lay cables because they're needed, without any silly market games.

The Bundespost would have you rent a slow, outdated modem and disallowed connecting any devices from the private market.

Source: this is was actually happened when they still had the monopoly.

It did, yes, but that policy could have easily been changed without privatization. Private companies often had the same policy ("Routerzwang") until the state stopped them. Whereas misaligned incentives are a direct consequence of trying to manage a natural monopoly using markets.

Here's an anecdotal experience:

1. I live in the UK, in a town 25 minutes by train from central London. The best broadband I can get is ~45Mb, and the best I've seen in the rest of the town is ~120Mb, provided by Virgin Broadband (which, for one reason or another, can't provide that service at my home).

2. I'm currently visiting my family in Croatia, and they've recently got a fibre connection; I just measured 527Mb [0]

[0] https://www.speedtest.net/result/10356498903

Tends to vary by building in London. Gigabit fiber is available in a decent portion of them. (Though tends to be more in the 700-800 range in practice)

Canada has the same problem. Prior to COVID-19, the banks I talked to really didn't like the risk profile of fibre builds. Real estate was the be all and end all. Want an interest only million dollar commercial real estate loan? Sure! Broadband infrastructure? Nope, that's too risky. As a small provider, you need to get to somewhere north of 500 customers just to break even, and then you need your next $5-10 million to keep building. Governments think that point in time funding programs are the solution, but they're not. Small communities need access to long term amortization of construction costs, similar to other government programs for large infrastructure projects that are available to municipalities. It's a real challenge to get anywhere in underserved communities.

How many users are using it though compared to incumbents. Once the load goes up they will need to deploy more and more satellites. However seems like they can do this at will now. This gives remote working outside urban areas an attractive potential.

Alternative headline: “ISP with hardly any users and therefore no congestion turns out to be really fast!”

Alternative title: "Brand new company with no existing market share successfully provides vital service that billion dollar companies with decades of experience failed to do"

I'm amazed that the latency appears so low. I used satellite broadband for years here in very rural England, latency was more like 1000ms!

LEO vs GEO, that is the main advantage of starlink

Geosynchronous (35,786km from surface) versus VLEO (550km from surface).

Basically the difference in altitude between your satellite and these satellites is almost three times the earth's diameter.

If you take the US to Europe round-trip, the latency is even faster then the cable bound connection, because the light in vacuum (space) travels around 1/3 faster then in fiber.

Only once they have inter satellite communication working.

> because the light in vacuum (space) travels around 1/3 faster then in fiber

Pretty sure the speed of light is a constant... You may be confusing the latency of the network framers/etc.

So, you are wrong. The speed of light is not constant. :) The speed of light depends on the medium, where the reference (or max speed) is the speed of light in vacuum. The speed of light in a diamond is only half of the speed in vacuum for example! This is why the latency of the low orbit satellites is lower then the cable bound connection.

IANAS but from what I understand, the speed is constant, but it appears to travel more slowly due to the atoms found along the way, which absorb and reemit the photon.

In the case of the fibers, the light just bounces on the walls of the tube, effectively lengthening the distance it has to travel.

It's like saying that I did not drive a 50 mph simply because I took a longer detour: I still drove 50 mph, but I did not go in a straight line.

> In the case of the fibers, the light just bounces on the walls of the tube, effectively lengthening the distance it has to travel.

This is incorrect, the reduced speed of light is due to the properties of the material through which the light is traveling. The material has in fact changed the speed of light.

The wave equation of light is dictated by the permittivity and permeability of the material through which it propagates. This is just a property of electromagnetic waves. Only the speed of light in a vacuum is a constant. Or more accurately, the permeability and permittivity of a vacuum.

Also fun: low latency fiber is made in space.

https://upward.issnationallab.org/the-race-to-manufacture-zb...

The index of refraction of materials is caused by differences of the speed of light in those materials. However, according to the theory of relativity, the speed of light in a _vacuum_ is constant in all reference frames.

There is a neat web page where you can see the speed of light in different materials:

https://micro.magnet.fsu.edu/primer/java/speedoflight/index.....

I'm very curious how the system will hold up when 10s of thousands are using it. I think it will be a while until that happens though.

While that bandwidth is impressive, I'd be interested to know what the expected packet-delays through the atmosphere are. The lowest orbit they plan to use is 340 km which translates to a minimum 2 ms delay. While the equipment will rarely align like that, even 10 ms is fine for most human use.

They're bidding in FCC Auction 904 (currently in progress), which requires an absolute maximum round-trip latency of 100ms. Given the FCC penalties they wouldn't be bidding if they weren't damn sure they would be significantly lower than that bar.

Details:

  https://auctiondata.fcc.gov/public/projects/auction904

Auction should conclude in 2-3 weeks.

People are getting 30-40ms ping times generally. Far lower than other satellite options, as you would expect.

"For now, Starlink internet cuts off for seconds at a time every couple of minutes" - wait, what?

So, does that mean 95% of internet users are facing even worse quality of service? or does it mean that given a very small number of participants so far, starlink is able to allocate more bandwidth to everyone?

I remember reading an analysis here on a previous posting about starlink that under load, it is going to have pretty limited throughput (on top of higher latency). I do not see how that will get solved.

> So, does that mean 95% of internet users are facing even worse quality of service?

The first sentence of the article makes it clear the 95% figure is in reference to speed.

…and only is reported by some users. Full sentence:

“Some of the participants in SpaceX’s public Starlink beta are reporting internet speeds higher than 95 percent of U.S. internet users.”

So the same could be said of regular providers? "Some users report speeds higher than 95% of the other users"...

This really needs more precise wording. What percentage of Starlink users (more than 5%?) and what's the speed difference?

When 4g/LTE was new, a coworker was an early adopter of whatever cellphone that was one of the first to support it. They were getting somewhere ~250 mbps on speed test.

Fast forward to everyone having 4g phones, now in practice I see like 1-2% of those speeds typically.

I have no idea the inner workings of starlink's protocol, but these aspects of wireless seem inevitable. once theres more simultaneous connections than in limited beta, each connection gets a smaller time slice or less frequency band, no?

I do not see how its solved either.

Currently I have rural DSL, approximately 3mbps down, 1 mbps up, but cheaper and reliable. The one person who has starlink beta is posting 200+ mbps screenshots but if everyone here connects I'm skeptical the result isn't slower, less reliable, more expensive than what's already here.

I can see that just over a day. During off-peak hours, I've speedtested around 260 Mbit/s on LTE on my iPhone (it's 10:20 PM on a weekday and I just got 230 Mbit down[0]), but during peak hours (commute hours or lunch rush), it drops to like... 20 Mbit

[0]https://www.speedtest.net/result/i/4226615772

IIRC initial LTE (UE category 3) only supports max 100Mbps (on spec) by 15MHz bandwidth. So I had measured max 50Mbps on 75Mbps (on spec, 10MHz) network.

"but those interruptions will likely fade as more satellites are launched and coverage becomes more comprehensive."

Coverage is incomplete as they’re not nearly done deploying satellites. Satellite TV comes from single satellites orbiting in geosynchronous orbit, way out there where the orbital period is exactly one day.

Starlink satellites are a swarm which operates in a low earth orbit with a period of around 90 minutes, you’re constantly changing satellites much like a cell phone switches towers as you drive.

A coverage gap would probably lead to multi-minute gaps in service.

Instead I'd guess this is a delay during reorientation of the physical antenna or repointing of the phased array.

The phased array can theoretically be repointed in milliseconds, but in a rush to launch I wouldn't be surprised if it doesn't take a few seconds to lock on to the signal, do channel characterization, and adjust data routing stuff to use the new link. During that time, a bunch of packets are lost, and you then have to wait for the TCP timers to expire before retransmission.

End user ends up seeing a 5 second glitch.

download speed is great, but what’s the latency like?

Seems low. https://www.reddit.com/r/Starlink/comments/jltvk8/way_out_in...

more info here: https://www.businessinsider.com/starlink-internet-satellite-...

30-40ms round trip generally. It's Starlink's biggest advantage vs. existing satellites, along with greater capacity per user due to the huge number of satellites, so they don't need draconian data caps.

They'll need some sort of congestion control in urban areas. They might just refuse to sell into urban areas, or use fair queuing, but urban users are going to be limited or capped one way or another.

I think Starlink has always marketed itself as a service provider for remote and hard to service areas. I'm not sure why people here keep thinking they will provide service to urban areas.

Sometimes you have not-great connection even deep in the city. I’m in a city touching Paris right now, 60k persons on a 1.5 by 1.5km square, one of the densest part of France.

Well I have a shit ADSL (not VDSL !) connection, despite everyone else in the street having FTTH, because the HOA of my building won’t bother…

On your situation, 4G/5G should be a solution rather than Starlink.

Reply on news.ycombinator.com