The problem with DOCSIS though is contention; not 'headline speeds'. I'm sure you could have ran a gigabit network on DOCSIS back then; but since that all networks have split the CMTS nodes more and more. If you enabled gigabit in 2013 on most networks it would have ground to a crawl because there was too many users per segment. Since then the networks have split the segment(fewer people competing for the same resources) and also the tech has got better with 3.1 and now 4.0. It's not quite as simple as flipping a switch and giving everyone gigabit.
A long time. I worked with Charter briefly 20 or so years ago, and the plan even then was to slowly roll up speeds, to always be giving people something to 'upgrade' to.
It’s amazing the subtle ways that pursuit of profit can slow down progress. We’re well aware of the big obvious ways that can happen but there’s seemingly small things like this too which add up.
I think in this case they are related. There are network effects to building the lines, so the difficulty with earning a profit prevents more companies from building out their network. This leads to a lack of competition (I’m sure government intervention has adversely affected this too).
But an organization like a co-op or a government would built the lines with no concern for profitability, and find some scheme to lease them to users at cost. Compared to a for-profit which artificially keeps speeds low, the government or co-op would lead to better cheaper service.
So broadly the pursuit of profit changes the way the organization works. The profit incentive potentially directly leads to the lack of competition. It also leads these organizations to lobby for government intervention which impedes competition. This last point is also why I would advocate for a co-op over government management, but I’d take either over for-profit networks.
> government would built the lines with no concern for profitability
>It also leads these organizations to lobby for government intervention which impedes competition
The problem with both of these scenarios is the same, government corruption. Without the corruption, municipal fiber or the free market would work great.
The bad guys here are the corrupt regulators, not the evil capitalists.
> The bad guys here are the corrupt regulators, not the evil capitalists.
Absolutely no reason to believe it must be one or the other. They are in fact working together. You propose that eliminating the government component would fix the issue, and I contend that eliminating the for-profit component would fix the issue. Both can be true.
> Without the corruption, municipal fiber or the free market would work great.
Entrenched networks are a well established phenomenon in relatively free markets. The cost to build a replacement for Facebook is relatively low, yet they still have a large moat due to their sizable subscriber base. In a physical system like a digital data network, the start up costs are much higher. It’s not a direct comparison to Facebook because one ISP is generally as good as any other, but generally it is difficult to compete with established companies when gaining a foothold involves a substantial investment from the outset.
Finally I did directly address your main point, I’m not quite sure why we’re still at the “either government or private for-profit firms” stage of the discussion when co-ops are a viable third option:
>> This last point is also why I would advocate for a co-op over government management, but I’d take either over for-profit networks.
>But an organization like a co-op or a government would built the lines with no concern for profitability
No, they are absolutely concerned about staying in the green. Apart from the federal government, local governments have to make enough money to avoid bankruptcy. Co-ops are even more constrained.
This makes them even less likely to invest in risky cutting edge technology or anything approaching R&D.
The only place these works are well established technologies. We’ve hit the point in the last decade where fiber is pretty well understood and future proof, so it now is low risk to lay a bunch of it and relatively cheap to upgrade endpoints as needed.
Make no mistake though, the only reason fiber has reached the commoditization point is because of a bunch of very profit driven networking gear manufactures dethroning each other over who can pack the most bits over wavelength.
that's exactly the same issue though. anybody who can potentially eat into the profits of the big ISPs simply gets bought up.
that's why the only hope is municipal fibre - we've all seen what happens to the new ISP who's going to save us from the greedy telcos, they disappear. municipal governments are at least a little bit resistant to being bought.
No, the issue is that nobody can compete because of NIMBYs and regulatory capture. Comcast doesn’t buy competitors, it just picks up regional cable providers in areas it doesn’t already own. Companies that try to compete without government intervention are doomed to fail.
According to Wikipedia, DOCSIS 4.0 was standardized in 2017 with 10gig down and 6gig up. But also DOCSIS 3.1 was from 2013 and could do "1-2 Gbps" upstream. So it seems like 1gig up, and maybe 2gig up, has been available for a decade, as far as the protocols.
It’s no surprise that one of the places they’re doing this, Colorado Springs, is pursuing municipal fiber optic. There’s no better way to overcome the complacency of your local cable monopoly than to publish a plan to offer a competitive service. They’ll improve their service and prices before you can even deploy the first strand of fiber.
google is in the business of intermediation. they really want to be the interface between you and the rest of the world. Fiber was just the best, most ultimate version of that - it was just really messy and expensive.
I think it served multiple purposes, they didn't raise consumer expectations at the time/ popularize 1 gig fiber service.
Around the same time the FTC was considering (and did) raise the speed definition of broadband and they YouTube had a big push to make 1080p streaming smooth and accessible, and they were being held back by a lot of slow last mile ISPs.
> So, now Google isn't quite as fussed about fibre.
Even if they were they couldn't get into a whole lot of markets. It sort of piggy backs off of municipal fiber which is a nonstarter in some places due to costs or it straight up being illegal.
Google also has Google Fi which is a mobile provider.
1) Google didn't want to be a service provider. ("Customer service? What's that??") It only needed to feign the threat to get the other real ISPs to pick up the speed. They were only committed enough to acheive that broader goal.
2) "Failure" is good for Google. It's a cover for where they are dominant. Shuttering Fiber isn't a failure. It's a win.
It’s a bug in Google’s promo driven culture. You don’t get promoted at Google by being good at your job, you have to show “impact”. That’s much easier to do by being on a new initiative. You don’t do that by improving and maintaining existing projects.
Amazon has the same “impact” driven promo culture. I haven’t worked at Google. But I did work at Amazon. From what I read about Google, it’s the same Hunger Games mentality.
Obviously not? Where are you living where even if fiber is available in your city, it’s not just a small section you’d need to deliberately account for in your selection of housing?
Minneapolis has a private fiber ISP, my brother is in Madison and they too have private fiber available widely. I was under the impression there had been a lot of fiber deployment in cities in the past five years.
US Internet (private fiber ISP in the Twin Cities) is an anomaly, not the norm. Minneapolis is incredibly lucky to have a local, private ISP that offers fiber to the home at up to 10 gigabit/s. The only other ISPs in the US that I know offer similar services are either municipal ISPs, co-ops, or Google Fiber.
Cincinnati, Ohio has Altafiber (formerly Cincinnati Bell) that offers 2gbps down 1gbps up for $80/mo. I have it and it’s amazing.
Cincinnati Bell was early to roll out DSL too. Went to college here in the late 90s and had a mind blowing 768kbps. I could watch RealVideo at 640x480!
The Midwest absolutely has seen a lot of fiber going in the ground with federal money the past couple years and Metronet has expanded pretty aggressively.
Not even close. In my research, only housing developments with new underground utilities since ~2005 to 2010, or sometimes anything super old with overhead wires.
Anywhere developed between 1980 to 2005 with underground utilities or anywhere with dense apartment buildings like NYC have effectively zero fiber.
Counter argument: my neighborhood was developed in the 1970s with underground utilities (and electric only, no natural gas, which is a bit odd in my part of the USA). They installed gas lines in the 90s (ripping up all of the streets) and fiber in the 10s, running alongside the street and using directional drilling to pass crossroads.
Popcorn ceilings are the opposite of "character." They'll give you cancer if you drill through them AND they're dated, ugly as sin, and impossible to keep clean.
Nope. In many areas yes but availability of service is still highly local. Even if you're near net the ISP could still want a build out fee vs an existing provider that's already on property.
It's pretty widespread in Seattle. We have Google Fiber, CenturyLink, and Wave G providing gigabit fiber services in different parts of the city. Then there is Atlas Networks which provides symmetrical gigabit services to apartment buildings via point to point wireless. I have Atlas and it has been fantastic.
I don't believe Google has fiber in Seattle. There is "GFiber" which from their site is: "GFiber is the company that offers Google Fiber and Google Fiber Webpass, high-speed broadband internet services that use fiber optic and wireless millimeter wave technology to deliver fast internet right to your home or business."
But millimeter wave wireless is available from Google in Seattle
Exactly. They didn’t do jack to improve their offerings until municipal fiber became a real threat. Now they are doing what they should have been doing for years: actually improving their product and competing. I hope it’s too little too late and people embrace municipal fiber!
The idea of a free market is that competition can enter the field. But, this has been an area that Issue have played every dirty angle possible to rig the game, including legislature and fees to setup the infrastructure. Now they're already settled in- there aren't any fees to pay for e.g. Comcast but new ISP Z is going to have a massive wall to overcome before seeing a penny back.
Maybe this is just my ignorance, but I honestly don’t care how they get these speeds to me. If I’m actually getting 2Gbps symmetrical, I don’t care if it’s because they installed a hamster wheel outside my house and that little guy is losing his mind making it happen. Let’s give credit where credit is due.
Exactly. Municipal broadband is a potentially existential threat to Comcast. Comcast is highly motivated to see early high-profile municipal projects fail. Each household they siphon off with this service drives up the average cost per subscriber of the municipal option. If they can get enough, the municipal option will turn into a giant money pit with a bunch of disgruntled taxpayers.
A few years down the road, when other cities are considering their options, someone will inevitably point to Colorado Springs. Comcast really, really, wants that data point to be a bad one.
I used to work on backend systems at both comcast and cox. Comcast’s engineering teams are stronk especially compared to Cox. One thing I took away from meeting their leadership team was that they were paying attention all of the questions an outsider was asking. The trouble that plagued them was set top boxes and they made an engineering decision over 10 years ago to git rid of them. Today I fully understand and appreciate how they are the number one cable isp.
Food for thought: they pay their engineers FAANG salaries, which explains the high service costs. Their technicians can make 100k+. The costs are high but you pay for what you get, my issue with that is you are forced into that price which most municipalities must pay
It's a competitive open access network - customers can pick which ISP they connect to. The services offered range from affordable ($80/mo gigabit) to... way beyond what most residential customers in the US can get ($280/mo for 10gbps service). https://xmission.com/utopia
Competition, it turns out, is awesome but sharing last-mile fiber resources is awesome too so that you can have competition without having to pay for multiple fiber links per house.
(And from my experience in Pittsburgh - Comcast's pricing is abysmal. Verizon FiOS is very reasonable but still more expensive with worse limits than Utopia. My almost-gigabit home service on FiOS is $90/month; my backup 20mbps Comcast link is $50/month.)
You have any evidence that that's the case? Everyone I know with access to municipal fiber loves it vs their private alternatives.
The competency of Comcast's network engineering department is irrelevant, they aren't in charge. They can only do what corporate is willing to fund, and corporate is happy to let its captive markets' infrastructure rot. It boosts their stock price to do so. If a municipality tried that, there could at least be consequences via the local government.
Perhaps the ideal is what we already do for every other utility, heavy regulation but managed by private entities. Of course, every time we try that the telecom lobbyists pout, make apocalyptic ads and throw money at congress critters until it stops. So municipal fiber is the workaround we have. If our choice is between two bad options, I'll take the option where I at least have a vote/can run for office and potentially change something vs the option where I'm just screwed.
Truth, but how many people must be in your household such that one would need 2 Gbps internet? For me, I think I'd need 20 people under my roof for this speed to have any chance of making any sense.
It is not about having a consistent need for High Speed , or in this case 2Gbps Internet, it is about there when you need it. Things like uploading or downloading games, movies, patches that could be done in seconds instead of sitting there waiting for minutes.
It’s also a case of not knowing what you’re missing until you have it, like getting glasses for the first time. I would accomplish so much less if I had to consider a 100mb speed limit.
> It’s also a case of not knowing what you’re missing until you have it, like getting glasses for the first time. I would accomplish so much less if I had to consider a 100mb speed limit.
Not really, I have access to fast internet and feel like I do just fine at 1/10th the speed.
I don’t know about you but other than me, no one in my house used substantial internet bandwidth. Netflix and YouTube mostly. If I throttled the house to 50mbps no one would notice.
For me, I pay for the highest speed because I know I'll never get there. If I paid for a 300mbps line I'd get maybe 10-20. If I pay for a 1.2gbps line guess what? I get about 500mbps. That's definitely more than enough for basically anything I do right now but lowering my speed to the lower tier will mean I'm invariably half way back toward 10mbps.
I agree with every word you said. You will probably need it only for 1% of the time.
But you see, 100Mbps Internet connection dont often give you 100Mbps connection all the time, if gives you a share of total connection, whether that is building, floor, flat, etc that guarantees under X condition you get 100Mbps. So in reality you really want this number to be at least 10X because the system was designed to be oversubscribed. ( And it is not the fault of any body, the tech was designed as such. ) And it is especially true for DOCSIS 4.0.
And in the case of Network planning, it will actually be better if they could speed up your 1% need so you leave more time and space for others. ( Of course some people abuse the network but that is a different story )
This argument sucks! Just because you can't figure out how to use it, doesnt mean there aren't 50,000,000 legit uses or future uses that nobody has thought of. Build it and they will come... If people had crazy fast network speeds at home, I guarantee you'd see services pop that could utilize them. As it is now, nobody can see past their app store and streaming video services, so we will just keep getting stuck with shitty upload speeds, and potential for new products and services stifled.
I had symmetrical gigabit fiber up and down via AT&T. Besides speed tests, the only times in the real world I ever remember getting close to those speeds were multithreaded BackBlaze backups, torrents, and connections to AWS for copying files back and forth to S3.
Some people work from home and do things like push Docker images or ML models up, especially over VPNs. Some other users create video content or work in creative spaces that deal with larger files. Comcast’s 45mbps aka 5.625MB/s limit on their highest tier 1200mbps residential service is absolutely a friction point for many of us today.
You’re going to fine your VPN is still going to be a limiting factor unless your company has really good bandwidth to support it.
Even with symmetrical 1Gbos and pushing to AWS, the VPN was slowing me down.
Luckily, we started implementing zero trust and not needing a VPN. But I would still just launch a Cloud9 instance and do everything on it for much faster speeds
And it did help that I worked for the only company that never has to worry about or pay AWS a huge AWS bill no matter how much we used it…
At that point you also need expensive networking hardware and SSDs to get that kind of speed. Most people stream movies rather than download so there it's okay even if it takes hours to download the whole content.
I think the house with many people case is overwhelmingly more common than the single networking enthusiast gamer/pirate.
(I have gigabit at home but probably wouldn't get 2 gigabit)
First, who doesn’t have SSDs in computers these days? Also, spinning rust drives haven’t been frozen in time since SSDs became mainstream. Modern 3.5” HDDs can hit >2000mbps, which is 250 MB/s. See eg https://www.tomshardware.com/reviews/seagate-firecuda-8tb-hd...
> expensive networking hardware
Many newer consumer-grade home routers have 2.5G ports, eg GL.iNet GL-MT3000 which is $100 right now https://a.co/d/bowJMRn
The important number here is not 2000mbps aka 250MBps. The important number is the current upload speed for the highest Comcast DOCSIS 3.1 “1200mbps” plan which is 45mbps aka 5.625 MB/s. To put that into perspective time-wise in networking terms, 802.11G was out on 2003 — 20 years ago — and that was rated at 54mbps. In other words, current Comcast aka XFINITY upload speeds could be saturated by wifi technology that is old enough to vote.
Restricting uploads ensures that you can't use any peer to peer applications to their full potential. In other words, it keeps you dependent on cloud services and other value bullshit add-ons. Why would I need cloud storage if I could just access my NAS when I'm out and about?
For me, it’s not just about the download, it’s about the upload and the data caps. My current 1200mbps down, 45mbps up Comcast plan, which is the fastest plan available from any ISP in our Silicon Valley home, has a 1.2TB data cap. If run at full speed, that data cap is exhausted in 2 hours, 13 minutes, and 20 seconds. Not even 2 and a quarter hours of downloading. For the “best” option. Lifting the data cap is another $30/month and they make you use their router hardware which opens up your network for anyone to use — that’s how “xfinitywifi” works, if you didn’t know…and since they then don’t want to attempt to account for the traffic difference between “your” internet usage and xfinitywifi traffic, you get unlimited data “free”. IIRC there used to be no unlimited option, and if you don’t have an unlimited plan, you get overage charges up to an additional $100 per month:
```
Once your courtesy month has been used, the next time you exceed 1.2 TB in a month (and any subsequent times it's exceeded in a month), you'll be notified by email and text message that you’ve exceeded your data usage plan amount. A block of 50GB will be added to your account, and a $10 charge will be applied to your bill automatically.
You'll be charged $10 each time we provide you with an additional block of 50GB of data, with a limit of $100 per month.```
The upload side would take 2 days, 11 hours, and 15 and a half minutes to upload 1.2TB. They do this to prevent torrenting and any meaningful file sharing, because “ NBCUniversal - Comcast Corporation - Xfinity” has an obvious incentive to protect their profits on the media side. DOCSIS channels are the commonly-cited technical reason. You can read more about those limits here https://en.m.wikipedia.org/wiki/DOCSIS#Throughput but the throughput caps could easily be 100mbps or 200mbps upload at my location with my current DOCSIS3.1 stuff.
Symmetrical 2gbps would be a dream, especially with unlimited data, but more than anything I’d love for these crooks to have some actual competition.
Good luck convincing every city in the US to rubber-stamp cutting up cement and closing roads, let alone all the private property owners that infra might have to go through. It’s some of the same reasons that we can’t build public transit.
The fact that even today in Silicon Valley, the fastest internet I can get at home is through Comcast with a 45mbps upload cap (1200mbps down) is a disgusting failure of every level of government to actually do right by consumers instead of just stuffing our tax dollars into the back pockets of giant companies while our infrastructure rots, and I mean that quite literally looking at the telephone poles down our streets here that still carry the coax used to deliver internet service. There’s no fiber in our suburban neighborhood because many of the residents bought their houses in the 60s and 70s, pay 1/20th or less of the property taxes that new home buyers do because of CA Prop 13 (which is also a big reason California cities always have budget issues — schools here that are surrounded by >$1m houses for miles go underfunded because of Prop 13), and aren’t willing to pay a dime more for anything like internet service because they’re already struggling to make ends meet. They’ve already been priced out of the neighborhood long ago, but we continue to pretend that giving them a $20k/yr tax break on their property tax is making the world better by letting them continue to be in poverty while living in a million-dollar+ property. I feel much less sorry for the wealthy folks who own many Prop 13 tax-advantaged properties here and rent them out at $8k/month in pure profit. I feel the least sorry for large companies like Disney that get CA Prop 13 benefits, but even stuff like 2020 CA Prop 15 fails at the ballots because everyone pearl clutches their sweet sweet tax break, even when we just try to limit it for large companies, because they think they’ll come for them next (see https://en.wikipedia.org/wiki/2020_California_Proposition_15 and https://calmatters.org/commentary/2020/08/how-prop-13-gave-c... )
I see our infrastructure problems as just another example of the older generations voting only out of self-interest with no desire to build a better society for their children, NIMBYism, and the ever-reliable culture of profits over people at every level o...
> I highly doubt we’re ever going to comprehensively fix all of our infrastructure issues in America.
We won't and we don't want to. People think that we're just a "few years out" from being able to fix things. We don't want to fix things. We love our country this way. This is what we vote for. This is what we pay for. This is us. We love being covered in the fetid scum at the bottom of a dumpster. It gets us off.
Coax is still going to run out of bandwidth eventually, so fiber lines run will not be wasted, but yeah, it is further out than most laypeople realized, going by other threads about coax, there's this popular conception that coax is super crappy, in reality it's a pretty good RF waveguide, unless it gets pierced or the connector is poorly crimped, but that also applies to fiber
Tbh using DOCSIS4.0 is probably a mistake and a dead end technology. Virgin Media in the UK said the cost of upgrading to 4.0 from their 3.1 network was only half that of actually laying full fibre to all their covered premises, so they are instead rolling out FTTH everywhere instead. 4.0 requires significant new physical network components deployed close to end users so it's not a cheap or easy upgrade.
Plus it's unlikely there will be much more speed after 4.0 given it is using very sophisticated modulation techniques and using nearly the entire spectrum of the coax cable.
That's interesting. I thought that, too, but recently I read that there's about another order of magnitude of bandwidth increase to squeeze out of coax yet. I didn't save the link though.
Sorry, I meant relative to fibre. I'm sure there will be DOCSIS 4.1 getting another 20% out of the plant, and various other small incremental improvements. Whereas really there is no reason you couldn't have terabit-PON systems.
> Tbh using DOCSIS4.0 is probably a mistake and a dead end technology. Virgin Media in the UK said the cost of upgrading to 4.0 from their 3.1 network was only half that of actually laying full fibre to all their covered premises, so they are instead rolling out FTTH everywhere instead.
Let's put this into proper perspective.
The state of Texas alone is ~2.8x larger than the entire UK.
Comcast FY22 Residential Broadband generated ~5x more revenue (spot FX; about the same circa end of CY22) and serves ~5x more customers [1; pp. 42-43] than all of Virgin Media O2 Fixed-Line over the same period [2; pp. 5-6].
To think what might make infra capex business sense for Virgin Media in the UK would be asymptotically transferable to Comcast in the US as a broadcast telecom with 6 decades of history and at the prevailing cost of capital strikes me as quite naive.
Average age of coax line in TX might be newer too, since the state has had a lot of new property development in recent decades.
I have a feeling that quality of the existing coax might be a factor in how successful or costly speed upgrades can be, since we're now squeezing the last bit of bandwidth out of them.
Cables that are out of spec but still mostly working at lower bandwidths are going to be a problem since higher bandwidth is more sensitive to the quality of the cable.
Some lines will have to be replaced, which will increase costs.
The issue is rarely the cables - coax run in the last 50 years is essentially identical to coax run now. It had to be to support all those TV channels!
Connectors do sometimes corrode, rodents do occasionally eat things, and they do need to upgrade some of their splitting/serving equipment though.
I'm well aware that Texas is bigger than the UK; and I'm also aware that Comcast is far bigger than VM.
I also said 'probably' a mistake, perhaps Comcast is right in doing this as maybe they have split the CMTSs aggressively and the cost of 4.0 is much lower for them than VM.
The other issue that VM is facing is energy costs which are much higher in the UK than the US. Coax requires a lot more power than (XS)GPON; which is pretty much all passive and therefore cheap to run.
Regardless, the long term future is going to be FTTH, not coax.
Who cares about the SIZE of Texas? Allow multiple fiber Carriers to operate in the state and build a statewide common infrastructure. Presuming there can be only one is limited thinking.
Consider Internet infrastructure to be a core state & national economic mainstay. Hold your governor and state legislators accountable.
Pressure your Federal representatives to do the same.
The USA has lost the will to all aspects of infrastructure, including power, high speed rail, next generation highways, bridges etc forward.
They’ll provide jobs for companies building the tech and for deployment.
First off, the article is about Comcast and what they're pushing on largely existing infra, while my criticism was directed towards a remark which struck me as uncritically dismissing the value prop of deployed tech just because it may make business sense to a much smaller private company with dramatically different capital requirements operating over a much smaller area in a completely different country; your entire remark is unconstructive and entirely irrelevant in this context.
To humor your rhetorical question on why size matters: When you're a publicly traded company operating in 40 states of varying population densities with over $55 billion in P&E assets and $93 billion in long-term debt (whose annual interest servicing requirement alone would consume all of Virgin Media O2 Fixed-Line revenue) on balance sheet[1] predominantly structured around copper infra as a legacy broadcast telecom with 60 years of history, you might start exercising critical restraint as an entitled consumer who thinks scrapping decades of capex investment for the latest shiny can just magically happen at the flip of some imaginary political switch and without substantial burden to the public in which it was intended to serve. Size matters because sustainably maintaining P&E depreciation while servicing a proverbial fuck ton of debt are real business expenses accountable to a market with waffling sentiments.
Comcast already has ~186k employees on payroll today, ~17% of which are in the UK/EU. Did you even give an ounce of consideration to how those existing jobs would be impacted in the face of imminent disruption?
Do you really think TAM will substantially increase just because FTTH is being offered at a fundamentally much higher minimum entry price for basic service? The FCC is motioning towards broadband for everyone, not just the latest bleeding-edge-nice-to-have-shiny for those with excess disposable income who can afford it.
Who do you think will really be paying the billions in capex debt for new shiny to replace that which still has meaningful life ahead of it just to stay afloat? You're fully entitled to be an irrational consumer who trades in a perfectly operable yet underwater vehicle to finance a new EV at high-as-a-kite interest rates, but I'd expect a lot more sustainability/value common sense from telecom executives with real skin in the game.
Building the required infra on debt is one thing, operating and maintaining it an entirely different beast; read Robichaux on Malone for a historical clue. Small ISPs are already crying a river[2] over BEAD grant capital accessibility because for any proposed project requiring $X to build out, of which 75% will be covered by federal grant dollars, just demonstrating the capacity to have 46% skin in the game is apparently too arduous. As much as I'd like to see small players step in to meaningfully challenge large incumbents, what I really don't want are hundreds of flake LLCs grifting the Treasury while pretending to deliver when the grass is green then conveniently jumping ship just when the going gets tough. As a consumer, real world viability as a sustainable business is just as important as the basic service to be provided.
Furthermore, when the FCC needs to step in and drop the hammer on the multitude of clowns who will try and inevitably fail, it'll be us taxpayers paying those administrative costs, further setting the stage for Comcast et al to acquire shiny assets and regain abandoned market share at steep clearing discount.
The issue was always mostly about monopoly power and not technical limitation. A hypothetical second coax network would achieve the same end result, but there’s no reason not to go with fiber for new installations today.
I know nothing of the space, could you go into the opex a bit? I naively assume that once the wire is in the ground, all of the maintenance is in powering routers + staff + recovery when a backhoe inadvertently cuts a line.
How do coax and fiber differentiate themselves on long-term costs?
Coax has powered equipment in the field. Power costs money (opex). The powered equipment sometimes needs replacing, probably much more often than GPON (opex to roll the truck, capex maybe to replace broken equipment)
>At the end of the day there was plenty of juice to squeeze from existing infra
That's the argument that POTS was making 20-30 years ago. But now it's just rotting in the ground in our area. In fact, just today I was driving by a telecom pedestal that some traffic had hit and was wondering if anybody even notices anymore. The next intersection there was one that was barely in better shape.
~100 years ago they built the telephone systems. ~50 years ago they built the cable systems. Sure, they can squeeze some more out of the cable, until they can't, at which point we're going to wish we had invested in some more modern infrastructure.
Surprised it’s actually symmetrical. Though I’ve never understood why, it’s always perplexed and irritated me that they typically offer upload bandwidth at a tenth of download. In the day and age of remote work, you’d think this would be illegal.
Its because DOCSIS has a limited number of frequency channels which can be assigned as either upstream/downstream and if they're comically over-subscribing their infrastructure you can offset that a bit by allocating most channels to downstream. I guess their thinking is that most users are going to be doing download heavy tasks and not upload, so it doesn't matter.
I think it's perhaps 1% preference and 99% realization of the reality.
Up: "Hey YouTube, I'd like to watch this video."
Down: Hundreds of MB of data from YouTube
Up: "Hey <streaming TV>, I'd like to binge watch <XYZ>"
Down: Hundreds of MB of data from the streaming service
Asymmetrical connections got more asymmetrical over time, limiting the utility of upload and shaping use of the internet (in most of the US) into more of a content portal or cable tv-like experience than it might otherwise be. I don't think it was purely an emergent phenomenon based on tech and consumer demand, but rather had a strategic element. I'll spare you all my tenuous reasoning, though, since I haven't got any real evidence that was the case.
Plenty of consumer features could exist, like sharing actual high bitrate streams of your desktop or camera to others or syncing large amounts of data to servers for backup/sharing.
You miss the point. Those features do exist, and technically savvy folks like us use them (and they work fine with my asymetric connect). But the vast majority of users want YouTube/TikTok/etc to stream without a glitch and couldn't care less about backups, and as far as "high bitrate streams", pretty sure most users are thinking "works good on my iPhone, is good enough".
Most of Comcast’s internet is exchanged through transit-free peering agreements, so only about 1% of Comcast’s outbound traffic requires them to pay for IP transit. And that traffic is mainly overseas, so Comcast is effectively a Tier 1 network in the US. Which means their customers’ outbound traffic doesn’t cost them anything.
For most last-mile ISPs, peering isn’t a revenue source. Small ISPs have to pay for peering and IP transit, and large ISPs have transit-free peering agreements with most, if not all, of the internet.
However, large ISPs that enjoy regional monopolies can refuse to upgrade connections with peered networks in an attempt to force the content providers sending traffic over the peered network to instead peer directly with the ISP.
This tactic is really only feasible for ISPs who have captive customers that are unable to switch to another ISP, as the negotiation process requires the ISP to allow the service it is providing to its actual customers to degrade to such an extent that the content provider is forced to peer directly with the ISP.
I think it's not so much that they care, just marketing.
If you have 1.2Gbps total bandwidth and you can split it as you wish, which do you think looks better to customers? 1Gbps (+200Mbps upload) or 600Mbps symmetric? Easier to sell a bigger number.
In my area it's more like 1Gbps down and 30Mbps up, which is very frustrating. That's only the tip of my iceberg of complaints about Comcast here. (A competition-free market, of course.)
Upload = people able to host and do things on their own.
Media companies like NBC/Sky/Universal/Xfinity/Comcast do not want you to be creators. They want you to consume consume consume... and it doesnt count against your bandwidth if you consume from THEM.
The short but unfulfilling answer is that this is because the DOCSIS standard historically has allocated a much broader frequency range for DL than UL. Unlike other forms of communication (like cell) that can use similar frequency ranges for transmit and receive, DOCSIS tends to slice something like < 70Mhz for UL and >70Mhz to 1Ghz for DL (I'm probably remembering the details incorrectly, trust Google over me!). Switching the frequency ranges often requires different circuitry and therefore different hardware.
I would guess that unlike fiber -- rarely saturated in a consumer context -- people have ~always wanted more than what cable can provide and thus the operators needed to be strategic about the allocation of bandwidth between DL and UL, hence the asymmetry.
I would say that cell signals mostly can't use the same frequencies for sending and receiving. The newer TDD allocations at 2.5GHz and above that carriers in the US are rolling out today do use the same for transmit and receive, but most of the old frequency ranges are exclusively up or down.
The difference is that cellular licenses generally offered equal bandwidth to up and down (ex. 10MHz down and 10MHz up). That's because they were originally designed for somewhat symmetrical communication: phone conversations where you assume each party is talking (sending) and listening (receiving) reasonably equally. With DOCSIS, it was designed assuming (correctly) that there'd be a lot more download traffic than upload traffic (which still holds true even in the era of remote work).
Actually most of the capacity problems on cable networks come from upstream congestion; not downstream congestion and this has nearly always been the case. When the upstream gets congested TCP ACKs get dropped and it kills download speed.
Upstream has always been challenging on coax. The first cable modems didn't even use the coax for upstream, instead it used a separate dial up modem.
I'm not entirely sure of the details but even after that DOCSIS has always had to use much lower frequency channels for upstream which are much more limited vs downstream which tends to use higher frequencies. Potentially because it was hard to get high frequency transmitting upstream low cost CPE devices back in the day?
It's about prioritizing downstream since that's 90% of the traffic and what customers complain about the most. No normal customer complains about their upstream cap, but they will freak out over congestion or not having the fastest download speeds
That, also the relatively low frequency range was originally only there for the sake of public access and the like; they only expected a relatively small number of channels on the return path.
I didn’t generally see upstream congestion when I worked on Internet QoE at Comcast, downstream was more common. Cable modems do ACK suppression as well to help there.
It's been a technical limitation. DOCSIS 3.1 and below allocate certain frequencies to download and certain frequencies to upload. There's a lot more allocated to download because of how people typically use the connection. DOCSIS 3.1 moves to a more flexible channel structure that allows for high-split systems. Xfinity has been rolling out 75-200Mbps upload speeds with their DOCSIS 3.1 network, but DOCSIS 3.1 is still lacking for gigabit symmetrical connections.
DOCSIS 4 mostly removes that limitation and allows for the full amount of spectrum to be used for up or down.
You need about 4Mbps up for a high quality stream up to converse live in HD. A reasonable asymmetrical connection on plain Jane docsis 3.1 already offers 20-50Mbps. It being asymmetrical isn't preventing many from working remotely.
A bigger issue is how many areas are so poorly served that in the present time there is no point calling it high speed.
That said ISPs tend to scale up with down so even if you might be reasonably served with 100Mbps down if you want a reasonable up you will have to upgrade to the higher down. If you intend to work from home this doesn't seem altogether unreasonable. When working for an ISP I was always surprised about how many complainers whose remote work was "so" important paid for bad speed, delivered via bad hardware, connected to bad wifi, on another floor.
> Even with good compression, video content still requires a significant amount of bandwidth. For example, YouTube recommends sending 34 Mbps for 4Kp30, and 50 Mbps for 4Kp60.
> In addition to needing high internet speed for 4k streaming, you need to factor in more bandwidth for your audio (typically in the 128–320 Kbps range) and a small amount more for overhead. Then double it — you should plan to always stay at 50% or less of available bandwidth in case of disruptions
Every major cable ISP in the US works this way. Logically this is because they aren't upstream limited except insofar as the inherent limits of DOCSIS 3.1 providing higher upstream is a very obvious market segmentation tool to get folks to pay for higher tier plans.
For instance here is a good example of Comcasts scale in upload
Please forgive the annoying second party sources most providers have this annoying flow where you have to almost sign up for internet to see specs and such.
Huh, okay. Well I couldn't find good info of increases last time I looked, and Spectrum was running 100 and 300 at the exact same upload. Good to know I could overpay for 35 if I needed to.
In addition to the frequency split mentioned by others, you can get much better utilization on the downstream channels than the upstream channels. Both are shared between all users on a node, but downstream has one transmitter and upstream has many, coordination and equipment tolerances mean the upstream can't be as tightly multiplexed. (GPON runs at 2.4G down, 1.2G up for this reason)
However, even with work from home, most users aren't pushing that much data. Teleconferencing is usually many streams in, one stream out. People working with large data feeds typically get a workstation somewhere with good connectivity and remote into it, where you send keystrokes and get big pictures. I have a pretty sad upstream, and it's not been a major problem, even if when I am pushing things from my home. Although there are a few tasks I avoided from home when I had a well connected office a few days a week.
For why it's symmetrical now specifically: killing QAM TV delivery frees up a lot of bandwidth for DOCSIS. Until that was a possibility, DOCSIS competed with 200+ channels for cable companies to allocate spectrum to.
I think US cable companies have to offer the option indefinitely because of CableCard, but they’re certainly moving to drop legacy TV delivery as much as possible.
Because actual traffic for the overwhelmingly vast majority of users is asymmetrical, and traffic engineering isn't about the 1 out of 10,000 customers who want to run a server. Even remote work is highly asymmetrical, at least looking at stats on the couple of tens of thousands of users we have WFH.
Illegal? Under what delusional definition of 'illegal' would you classify "doing what works for pretty much everyone but you"?
This may surprise you, but the majority of broadband infrastructure in this country was set up before the pandemic and the WFH explosion. When 99 44/100 percent of people consumed exponentially more bandwidth downstream than upstream.
The reason cable is asymmetrical is because cable plants weren't designed for two-way communication. Cable originally was a one way broadcast system with a tiny upstream path they eventually leveraged for Video-On-Demand/Pay-per-view (VoD/PPV).
I used to work with cable modems, and it's miraculous how they got this huge amount of performance out of their physical plant by basically hacking on their stuff until it worked. The cable industry has this thin layer of really smart guys and a huge number of porn hound meatheads.
I mean, did you know cable internet is basically token ring? Seriously. Not sure if that's for all implementations or just Comcast.
These wires were never meant to do anything close to this...but here we are.
One of the reasons this is so hard BTW is that the plant has to also support all that legacy equipment that's still connected and running. There are people with cable that haven't upgraded their box for 50 years and never will. You don't want to impact their service, period. No calls, no truck rolls.
> You don't want to impact their service, period. No calls, no truck rolls.
From my experience this is absolutely not true. Comcast had zero issues intentionally crippling my internet connectivity because the modem I owned only supported DOCSIS 2.
I received letters for months that I needed to upgrade my modem to support a “the best speeds”. Then I didn’t, and my speed dropped to less than a megabit per second.
I could have waived it off as a coincidence, except a couple of months later I was telling the story at a happy hour and it turns out the exact same thing happened to another guy right around the same time.
Comcast does not care if you have a shitty experience. They don’t care if you call. Me not upgrading and calling support about it is just the cost of them doing business.
The ACTUAL answer is that it separates out business from residential service - and pricing.
When I first got cable decades ago, it was symmetrical by default.
But they figured out business customers need symmetrical and are price insensitive, while residential customers really just need consumption and are very price sensitive.
The idea is that they can charge businesses more, and individuals less and get away with it.
This is just like airfares, where business customers travel during the week and want to be home for the weekend, and individuals will stay the weekend and are price sensitive.
In the US, ISPs are usually pretty good about offering what they promise in terms of speeds. It is shared, but basically everything on the internet is shared. Even if you have fiber to your house, that fiber is ending up in a central office somewhere that then only has a certain amount of bandwidth to the rest of the internet.
ISPs like Xfinity are splitting nodes so that you aren't sharing the link with many people before you're on fiber and oversell ratios are pretty well understood at this point (and apply to all ISPs whether the wire coming to your house is shared or it just becomes shared farther upstream).
A municipal broadband network in my area has a 2Gbps connection to the wider internet and offers 1Gbps service to people's homes/businesses and that 2Gbps serves several hundreds. Clearly, not everyone could get gigabit service off that at the same time, but people's traffic is typically bursty.
The internet itself is shared at different points even if the wire to your house isn't shared. If you have dedicated fiber running from the central office to your house and then the central office only has X Gbps to the rest of the internet, you're still on a shared link regardless of how dedicated things are between you and the central office.
I'm on an oversaturated node. A field tech was able to confirm it for me after only being able to get about 800 down/25 up on my 1200/40 plan. I asked if they were planning on splitting it and they said they won't until they move the the new Harmonic nodes they're in the process of rolling out. Pretty frustrating.
My n=1 but yes, it's impeccable. I have Comcast cable internet + TV. Perhaps 10-15 years ago it went out whenever there was bad weather and remained down even after power had been restored after an outage.
I can't remember the last time it went out (outside of power outages). At least 10 years.
Boston area here, and genuinely can't complain for the past 15 years over 3 different cities or parts of Boston.
About 6 years ago I had some spottiness when I had moved, but a call had a senior tech rolled out, it was identified as too strong a signal as I was so close to the head-end in the building, put an attenuator on the connection to get it smack in the middle of its target range and it was perfect after that.
Even right now my bill has dropped from $90 to $80, and again recently from $80 to $75 with no contract, for 1.2Gbps down, 35Mbps up. In which I actually get the full 120% over-provisioned speed all the time. No bandwidth cap in Massachusetts, either. I run my own DNS servers in my homelab, and with the network equipment on UPS backup, even power outages have failed to cause a service interruption in the 4 years at my current location.
My area is going to soon start piloting its municipal fiber rollout soon. That I'm looking forward to as it'll give another option and increase competition. I look forward to having improved options -- likely at lower prices still, since the cities are targeting 1Gbps symmetrical for $50/mo -- one way or the other in the coming year or so.
Not for me. Seems to have short outages every week or two. Had about two weeks recently during which, about half the time, the outages were severe enough to show up on Comcast’s tracker.
Coax that's already been run to the home and pulled through walls is also "almost" free. Perhaps even moreso.
I switched to fiber at my current home the moment it became available, but until then I enjoyed being fairly well-served by gigabit (down) cable over DOCSIS 3.1 on the infrastructure that was ready to use on move-in day.
You seem to be missing that the biggest expense is actually running and burying the wires themselves. Coax is already deployed in huge parts of the US, and it’s not easy or fast to rewire the whole country with something new.
No reason a state couldn't decide to. New Zealand (approximately equal to Oregon) decided to do it in 2011 and finished the planned fibre rollout in about a decade.
This was a public-private partnership (funded by government but not rolled-out nor owned by our government).
The existing cable in my city is now unused (not sure how their incentives were structured, but I couldn't sign up to cable 4 years ago even though house was already wired for it)
I originally thought it was a dumb idea - state funding for netflix - however covid showed the value of the fibre connections.
It's kind of rare that I find sites able to actually push and pull at truly gigabit speeds. For a normal home user, today, even gigabit seems over the top.
I do think we'll look back in a few years and wonder how we got by with just 1 gig connections. Just saying today five gig home internet feels extraordinary extreme.
AT&T will offer me "Basic 5", Speeds "up to" 5MB/s. This is a great improvement over the shitty ADSL I used to have from AT&T, but I switched to Comcast and now have 800MB/s, and could upgrade to 1GB/s.
Yep. I’m one of those neighborhoods that just got Fiber installed from someone else. Of course the end sell from Comcast was to match the price, speed and assure me a 20% discount if I signed for 2 years. Oh, and they have fiber to home now, in my area today, just come to a xfinity store and get my ONT!!!. Of course today… last year you tried to lock me in a cable and tv package that also required phone for $250 a month.
Interesting, how many seconds of 2Gbps transfer do I get before I reach my monthly cap and they start throttling me?
Jokes aside, I'm curious how this is even possible over decades-old cable. I get there's a new DOCSIS standard, but I'm less interested in the protocol and more interested in the simple physics of it. How can a simple coaxial cable cram so much bandwidth?
Happy ex-Comcast customer as well (as much as it pains me to say it, ATT has actually been pretty good to me with their fiber) but your numbers seem to go against a 1 TB limit but even the gigabit pro plan already includes the "unlimited data" option which allows you to go well past that. They never would say what exactly "fair use" was but it was at least above 15 TB/m from what I could tell of not getting kicked off.
Oh interesting. When I was in California, my 1 Gbps plan came with only 1 TB of bandwidth. They did sell "unlimited" as an additional upgrade but IIRC it was $50/month.
> How can a simple coaxial cable cram so much bandwidth?
A large amount of spectrum to work with and a high spectral efficiency. Wikipedia lists DOCSIS 4.0 as having 1.8GHz of bandwidth, and DOCSIS 3.1 as ~10bits/Hz. Assuming DOCSIS 4.0 is as least as efficient, thats about 18Gbps.
Spectrum is nearly managing it as well. There is FTTN in my neighborhood but the connection to my residence is still coaxial and using a cable modem. At inverse-peak traffic times I am able to reach speeds up to around 1.4gbps between my homelab and the backup server I have at the colo across town.
I haven’t actually checked to see if the limiting factor is my own network hardware but it very well may be.
Comcast Business limits you to their slowest 50/15 Mbps plan if you use your own modem and they want $20/mo for ipv6. I hope the people making those decisions die painfully.
Comcast is one of the most disingenuous companies I know. They have been advertising their “10g” network and more reliable home internet even though 10g doesn’t matter at all for the vast majority of users paying for 5mbps up and 50mbps down.
Reliable internet for most is based on the quality of home WiFi network, not bandwidth or theoretical backend network that Comcast has.
I mean, Docsis 4.0 does 10 gbit down 6gbit up through old coax.
Apparently we can keep putting better and better signal processing on the 2 ends to interpret higher and higher bandwidth signals.
Charter Spectrum says:
Together, about 85% of Charter's footprint will be capable of offering up to 5 Gbit/s by the end of 2025 and the ability to offer up to 10 Gbit/s in some of its footprint.
10/6 gigabit is pretty theoretical. Depends on lengths of the segments, quality of cable and taps, etc., and after those kind of penalties the bandwidth is still shared between everyone on the node.
Went from Comcast (supposed) 900/20 Mbps connection to Google Fiber 2.0/1.0 Gbps connection. Unlike Comcast, I actually get the advertised speed with Google Fiber. I would frequently get no more than 70 Mbps down with Comcast. Data caps weren't a concern because their connection was too slow to be a problem. Cancelling my Comcast account took forever as they kept giving me the run around.
I don’t believe this will be widely rolled out. 2gb/200mbps down/up has supposedly been offered in my area for a year but I’ve never been able to get it. I switched to ATT fiber when it became available
Meanwhile, here in Australia the fastest widely available upload speed on our national network (NBN) is 40bbps (~5MB/s), with some lucky few able to get 50mbps (~6MB/s).
Seems like the network designers want the populace to be consumers rather than creators. :(
This is kind of funny, because in my home country, brasil, I hear most companies are in a frenzy to deploy fiber, because of widespread copper cable theft in some regions.
And meanwhile here in the US, Comcast will stubbornly be attached to their coax tables until the end of times.
213 comments
[ 3.2 ms ] story [ 244 ms ] threadHigh-split DOCSIS 3.1 topped out at about 1.5gbit/sec upstream shared among all customers on a segment.
DOCSIS 4.0 spec[1] was first issued circa Aug 2019 and has undergone 6-10 revisions since.
[1] https://www.cablelabs.com/specifications/search?currentPage=...
"up to 10 Gbit/s downstream and 1 Gbit/s upstream"
Most of the 4.0 improvement is upload.
3.1 wouldn't do 2Gbps symmetrical, but it would do 2Gbps down with a reasonable up.
But an organization like a co-op or a government would built the lines with no concern for profitability, and find some scheme to lease them to users at cost. Compared to a for-profit which artificially keeps speeds low, the government or co-op would lead to better cheaper service.
So broadly the pursuit of profit changes the way the organization works. The profit incentive potentially directly leads to the lack of competition. It also leads these organizations to lobby for government intervention which impedes competition. This last point is also why I would advocate for a co-op over government management, but I’d take either over for-profit networks.
>It also leads these organizations to lobby for government intervention which impedes competition
The problem with both of these scenarios is the same, government corruption. Without the corruption, municipal fiber or the free market would work great.
The bad guys here are the corrupt regulators, not the evil capitalists.
Absolutely no reason to believe it must be one or the other. They are in fact working together. You propose that eliminating the government component would fix the issue, and I contend that eliminating the for-profit component would fix the issue. Both can be true.
> Without the corruption, municipal fiber or the free market would work great.
Entrenched networks are a well established phenomenon in relatively free markets. The cost to build a replacement for Facebook is relatively low, yet they still have a large moat due to their sizable subscriber base. In a physical system like a digital data network, the start up costs are much higher. It’s not a direct comparison to Facebook because one ISP is generally as good as any other, but generally it is difficult to compete with established companies when gaining a foothold involves a substantial investment from the outset.
Finally I did directly address your main point, I’m not quite sure why we’re still at the “either government or private for-profit firms” stage of the discussion when co-ops are a viable third option:
>> This last point is also why I would advocate for a co-op over government management, but I’d take either over for-profit networks.
No, they are absolutely concerned about staying in the green. Apart from the federal government, local governments have to make enough money to avoid bankruptcy. Co-ops are even more constrained.
This makes them even less likely to invest in risky cutting edge technology or anything approaching R&D.
The only place these works are well established technologies. We’ve hit the point in the last decade where fiber is pretty well understood and future proof, so it now is low risk to lay a bunch of it and relatively cheap to upgrade endpoints as needed.
Make no mistake though, the only reason fiber has reached the commoditization point is because of a bunch of very profit driven networking gear manufactures dethroning each other over who can pack the most bits over wavelength.
that's why the only hope is municipal fibre - we've all seen what happens to the new ISP who's going to save us from the greedy telcos, they disappear. municipal governments are at least a little bit resistant to being bought.
No, the issue is that nobody can compete because of NIMBYs and regulatory capture. Comcast doesn’t buy competitors, it just picks up regional cable providers in areas it doesn’t already own. Companies that try to compete without government intervention are doomed to fail.
Around the same time the FTC was considering (and did) raise the speed definition of broadband and they YouTube had a big push to make 1080p streaming smooth and accessible, and they were being held back by a lot of slow last mile ISPs.
Then everybody went mobile. So, now Google isn't quite as fussed about fibre.
Even if they were they couldn't get into a whole lot of markets. It sort of piggy backs off of municipal fiber which is a nonstarter in some places due to costs or it straight up being illegal.
Google also has Google Fi which is a mobile provider.
https://arstechnica.com/information-technology/2019/02/googl...
1) Google didn't want to be a service provider. ("Customer service? What's that??") It only needed to feign the threat to get the other real ISPs to pick up the speed. They were only committed enough to acheive that broader goal.
2) "Failure" is good for Google. It's a cover for where they are dominant. Shuttering Fiber isn't a failure. It's a win.
It’s a bug in Google’s promo driven culture. You don’t get promoted at Google by being good at your job, you have to show “impact”. That’s much easier to do by being on a new initiative. You don’t do that by improving and maintaining existing projects.
Amazon has the same “impact” driven promo culture. I haven’t worked at Google. But I did work at Amazon. From what I read about Google, it’s the same Hunger Games mentality.
Cincinnati Bell was early to roll out DSL too. Went to college here in the late 90s and had a mind blowing 768kbps. I could watch RealVideo at 640x480!
Anywhere developed between 1980 to 2005 with underground utilities or anywhere with dense apartment buildings like NYC have effectively zero fiber.
I do agree they're a pain to clean/paint, but I'm saying there are still good use-cases for a textured ceiling.
The only alternative is to get something wireless, either a microwave antenna service or a cell-based hotspot
It doesn’t seem any companies are trying to compete or even improve their services
Spectrum goes down at least 3 times a month, sometimes for hours on end
I can get cable with fast downloads and 11Mbps uploads, or if I remember right the phone line option was 50Mbps each way.
But millimeter wave wireless is available from Google in Seattle
That would be the best scenario
I don’t even want to begin to think about how much power is needed to achieve that while minimizing noise etc.
A few years down the road, when other cities are considering their options, someone will inevitably point to Colorado Springs. Comcast really, really, wants that data point to be a bad one.
Food for thought: they pay their engineers FAANG salaries, which explains the high service costs. Their technicians can make 100k+. The costs are high but you pay for what you get, my issue with that is you are forced into that price which most municipalities must pay
Frankly, your description of consistent incompetence reads much more like my local cable internet monopoly…
UTOPIA in Utah is a good example of a major municipal fiber project.
https://en.m.wikipedia.org/wiki/Utah_Telecommunication_Open_...
It's a competitive open access network - customers can pick which ISP they connect to. The services offered range from affordable ($80/mo gigabit) to... way beyond what most residential customers in the US can get ($280/mo for 10gbps service). https://xmission.com/utopia
Competition, it turns out, is awesome but sharing last-mile fiber resources is awesome too so that you can have competition without having to pay for multiple fiber links per house.
(And from my experience in Pittsburgh - Comcast's pricing is abysmal. Verizon FiOS is very reasonable but still more expensive with worse limits than Utopia. My almost-gigabit home service on FiOS is $90/month; my backup 20mbps Comcast link is $50/month.)
The competency of Comcast's network engineering department is irrelevant, they aren't in charge. They can only do what corporate is willing to fund, and corporate is happy to let its captive markets' infrastructure rot. It boosts their stock price to do so. If a municipality tried that, there could at least be consequences via the local government.
Perhaps the ideal is what we already do for every other utility, heavy regulation but managed by private entities. Of course, every time we try that the telecom lobbyists pout, make apocalyptic ads and throw money at congress critters until it stops. So municipal fiber is the workaround we have. If our choice is between two bad options, I'll take the option where I at least have a vote/can run for office and potentially change something vs the option where I'm just screwed.
I choose my Comcast tier based on the upload speed, because that's the real bottleneck in the system. Making DOCSIS (almost) symmetric is huge.
I was on a ~200 Mbps connection for a while and to be honest, a 1 GB file takes less than a minute, which is crazy fast.
I get that things will change in the future, but right now, I'd say 98% of the population would be happy with a 100 Mbps connection.
Not really, I have access to fast internet and feel like I do just fine at 1/10th the speed.
But you see, 100Mbps Internet connection dont often give you 100Mbps connection all the time, if gives you a share of total connection, whether that is building, floor, flat, etc that guarantees under X condition you get 100Mbps. So in reality you really want this number to be at least 10X because the system was designed to be oversubscribed. ( And it is not the fault of any body, the tech was designed as such. ) And it is especially true for DOCSIS 4.0.
And in the case of Network planning, it will actually be better if they could speed up your 1% need so you leave more time and space for others. ( Of course some people abuse the network but that is a different story )
“Now you can hit your bandwidth cap in half the time!”
— Comcast
Then soon your site will be the slowest one around and you’d want to catch up.
Even with symmetrical 1Gbos and pushing to AWS, the VPN was slowing me down.
Luckily, we started implementing zero trust and not needing a VPN. But I would still just launch a Cloud9 instance and do everything on it for much faster speeds
And it did help that I worked for the only company that never has to worry about or pay AWS a huge AWS bill no matter how much we used it…
I think the house with many people case is overwhelmingly more common than the single networking enthusiast gamer/pirate.
(I have gigabit at home but probably wouldn't get 2 gigabit)
First, who doesn’t have SSDs in computers these days? Also, spinning rust drives haven’t been frozen in time since SSDs became mainstream. Modern 3.5” HDDs can hit >2000mbps, which is 250 MB/s. See eg https://www.tomshardware.com/reviews/seagate-firecuda-8tb-hd...
> expensive networking hardware
Many newer consumer-grade home routers have 2.5G ports, eg GL.iNet GL-MT3000 which is $100 right now https://a.co/d/bowJMRn
The important number here is not 2000mbps aka 250MBps. The important number is the current upload speed for the highest Comcast DOCSIS 3.1 “1200mbps” plan which is 45mbps aka 5.625 MB/s. To put that into perspective time-wise in networking terms, 802.11G was out on 2003 — 20 years ago — and that was rated at 54mbps. In other words, current Comcast aka XFINITY upload speeds could be saturated by wifi technology that is old enough to vote.
This only has one 2.5GB port so it's only suitable for the many-users case, not the downloading-stuff-fast-alone case.
A t-1 used to be enough for 25 people officially.
``` Once your courtesy month has been used, the next time you exceed 1.2 TB in a month (and any subsequent times it's exceeded in a month), you'll be notified by email and text message that you’ve exceeded your data usage plan amount. A block of 50GB will be added to your account, and a $10 charge will be applied to your bill automatically.
You'll be charged $10 each time we provide you with an additional block of 50GB of data, with a limit of $100 per month.```
https://www.xfinity.com/support/articles/data-usage-exceed-u...
The upload side would take 2 days, 11 hours, and 15 and a half minutes to upload 1.2TB. They do this to prevent torrenting and any meaningful file sharing, because “ NBCUniversal - Comcast Corporation - Xfinity” has an obvious incentive to protect their profits on the media side. DOCSIS channels are the commonly-cited technical reason. You can read more about those limits here https://en.m.wikipedia.org/wiki/DOCSIS#Throughput but the throughput caps could easily be 100mbps or 200mbps upload at my location with my current DOCSIS3.1 stuff.
Symmetrical 2gbps would be a dream, especially with unlimited data, but more than anything I’d love for these crooks to have some actual competition.
Cookie-cutter approvals and proposals, just enough to get the local monopolies to get off their backsides...
See https://www.pewtrusts.org/en/research-and-analysis/fact-shee... for a quick read summary of the issues.
See https://broadbandnow.com/report/municipal-broadband-roadbloc... for some of the legal chicanery that incumbent monipoly ISPs have put in place.
See https://www.wired.com/2013/07/we-need-to-stop-focusing-on-ju... for other local government struggles.
See https://www.vice.com/en/article/ywkn4b/study-throwing-taxpay... for some history on how these monopoly ISPs rob taxpayers without actually delivering.
The fact that even today in Silicon Valley, the fastest internet I can get at home is through Comcast with a 45mbps upload cap (1200mbps down) is a disgusting failure of every level of government to actually do right by consumers instead of just stuffing our tax dollars into the back pockets of giant companies while our infrastructure rots, and I mean that quite literally looking at the telephone poles down our streets here that still carry the coax used to deliver internet service. There’s no fiber in our suburban neighborhood because many of the residents bought their houses in the 60s and 70s, pay 1/20th or less of the property taxes that new home buyers do because of CA Prop 13 (which is also a big reason California cities always have budget issues — schools here that are surrounded by >$1m houses for miles go underfunded because of Prop 13), and aren’t willing to pay a dime more for anything like internet service because they’re already struggling to make ends meet. They’ve already been priced out of the neighborhood long ago, but we continue to pretend that giving them a $20k/yr tax break on their property tax is making the world better by letting them continue to be in poverty while living in a million-dollar+ property. I feel much less sorry for the wealthy folks who own many Prop 13 tax-advantaged properties here and rent them out at $8k/month in pure profit. I feel the least sorry for large companies like Disney that get CA Prop 13 benefits, but even stuff like 2020 CA Prop 15 fails at the ballots because everyone pearl clutches their sweet sweet tax break, even when we just try to limit it for large companies, because they think they’ll come for them next (see https://en.wikipedia.org/wiki/2020_California_Proposition_15 and https://calmatters.org/commentary/2020/08/how-prop-13-gave-c... )
I see our infrastructure problems as just another example of the older generations voting only out of self-interest with no desire to build a better society for their children, NIMBYism, and the ever-reliable culture of profits over people at every level o...
We won't and we don't want to. People think that we're just a "few years out" from being able to fix things. We don't want to fix things. We love our country this way. This is what we vote for. This is what we pay for. This is us. We love being covered in the fetid scum at the bottom of a dumpster. It gets us off.
At the end of the day there was plenty of juice to squeeze from existing infra
Plus it's unlikely there will be much more speed after 4.0 given it is using very sophisticated modulation techniques and using nearly the entire spectrum of the coax cable.
Let's put this into proper perspective.
The state of Texas alone is ~2.8x larger than the entire UK.
Comcast FY22 Residential Broadband generated ~5x more revenue (spot FX; about the same circa end of CY22) and serves ~5x more customers [1; pp. 42-43] than all of Virgin Media O2 Fixed-Line over the same period [2; pp. 5-6].
To think what might make infra capex business sense for Virgin Media in the UK would be asymptotically transferable to Comcast in the US as a broadcast telecom with 6 decades of history and at the prevailing cost of capital strikes me as quite naive.
[1] https://www.sec.gov/Archives/edgar/data/1166691/000116669123...
[2] https://news.virginmediao2.co.uk/wp-content/uploads/2023/02/...
I have a feeling that quality of the existing coax might be a factor in how successful or costly speed upgrades can be, since we're now squeezing the last bit of bandwidth out of them.
Cables that are out of spec but still mostly working at lower bandwidths are going to be a problem since higher bandwidth is more sensitive to the quality of the cable.
Some lines will have to be replaced, which will increase costs.
Connectors do sometimes corrode, rodents do occasionally eat things, and they do need to upgrade some of their splitting/serving equipment though.
I also said 'probably' a mistake, perhaps Comcast is right in doing this as maybe they have split the CMTSs aggressively and the cost of 4.0 is much lower for them than VM.
The other issue that VM is facing is energy costs which are much higher in the UK than the US. Coax requires a lot more power than (XS)GPON; which is pretty much all passive and therefore cheap to run.
Regardless, the long term future is going to be FTTH, not coax.
Consider Internet infrastructure to be a core state & national economic mainstay. Hold your governor and state legislators accountable.
Pressure your Federal representatives to do the same.
The USA has lost the will to all aspects of infrastructure, including power, high speed rail, next generation highways, bridges etc forward.
They’ll provide jobs for companies building the tech and for deployment.
To humor your rhetorical question on why size matters: When you're a publicly traded company operating in 40 states of varying population densities with over $55 billion in P&E assets and $93 billion in long-term debt (whose annual interest servicing requirement alone would consume all of Virgin Media O2 Fixed-Line revenue) on balance sheet[1] predominantly structured around copper infra as a legacy broadcast telecom with 60 years of history, you might start exercising critical restraint as an entitled consumer who thinks scrapping decades of capex investment for the latest shiny can just magically happen at the flip of some imaginary political switch and without substantial burden to the public in which it was intended to serve. Size matters because sustainably maintaining P&E depreciation while servicing a proverbial fuck ton of debt are real business expenses accountable to a market with waffling sentiments.
Comcast already has ~186k employees on payroll today, ~17% of which are in the UK/EU. Did you even give an ounce of consideration to how those existing jobs would be impacted in the face of imminent disruption?
Do you really think TAM will substantially increase just because FTTH is being offered at a fundamentally much higher minimum entry price for basic service? The FCC is motioning towards broadband for everyone, not just the latest bleeding-edge-nice-to-have-shiny for those with excess disposable income who can afford it.
Who do you think will really be paying the billions in capex debt for new shiny to replace that which still has meaningful life ahead of it just to stay afloat? You're fully entitled to be an irrational consumer who trades in a perfectly operable yet underwater vehicle to finance a new EV at high-as-a-kite interest rates, but I'd expect a lot more sustainability/value common sense from telecom executives with real skin in the game.
Building the required infra on debt is one thing, operating and maintaining it an entirely different beast; read Robichaux on Malone for a historical clue. Small ISPs are already crying a river[2] over BEAD grant capital accessibility because for any proposed project requiring $X to build out, of which 75% will be covered by federal grant dollars, just demonstrating the capacity to have 46% skin in the game is apparently too arduous. As much as I'd like to see small players step in to meaningfully challenge large incumbents, what I really don't want are hundreds of flake LLCs grifting the Treasury while pretending to deliver when the grass is green then conveniently jumping ship just when the going gets tough. As a consumer, real world viability as a sustainable business is just as important as the basic service to be provided.
Furthermore, when the FCC needs to step in and drop the hammer on the multitude of clowns who will try and inevitably fail, it'll be us taxpayers paying those administrative costs, further setting the stage for Comcast et al to acquire shiny assets and regain abandoned market share at steep clearing discount.
[1] https://www.sec.gov/Archives/edgar/data/1166691/000116669123...
[2] drozycki ↗ The issue was always mostly about monopoly power and not technical limitation. A hypothetical second coax network would achieve the same end result, but there’s no reason not to go with fiber for new installations today. gowings97 ↗ by time you go n+0 with all the nodes to enable DOCSIS 4.0, you probably could just do FTTH for nearly the same cost. fbdab103 ↗ I know nothing of the space, could you go into the opex a bit? I naively assume that once the wire is in the ground, all of the maintenance is in powering routers + staff + recovery when a backhoe inadvertently cuts a line. toast0 ↗ Coax has powered equipment in the field. Power costs money (opex). The powered equipment sometimes needs replacing, probably much more often than GPON (opex to roll the truck, capex maybe to replace broken equipment) gowings97 ↗ Precisely linsomniac ↗ >At the end of the day there was plenty of juice to squeeze from existing infra
FTTH opex is also lower after 5 years.
How do coax and fiber differentiate themselves on long-term costs?
That's the argument that POTS was making 20-30 years ago. But now it's just rotting in the ground in our area. In fact, just today I was driving by a telecom pedestal that some traffic had hit and was wondering if anybody even notices anymore. The next intersection there was one that was barely in better shape.
~100 years ago they built the telephone systems. ~50 years ago they built the cable systems. Sure, they can squeeze some more out of the cable, until they can't, at which point we're going to wish we had invested in some more modern infrastructure.
Summarizing greatly, but user upload = Comcast pays vs user download = Comcast's peers pay them.
I'd assume for last-mile ISPs, peering is a non-negligible source of revenue.
For most last-mile ISPs, peering isn’t a revenue source. Small ISPs have to pay for peering and IP transit, and large ISPs have transit-free peering agreements with most, if not all, of the internet.
However, large ISPs that enjoy regional monopolies can refuse to upgrade connections with peered networks in an attempt to force the content providers sending traffic over the peered network to instead peer directly with the ISP.
This tactic is really only feasible for ISPs who have captive customers that are unable to switch to another ISP, as the negotiation process requires the ISP to allow the service it is providing to its actual customers to degrade to such an extent that the content provider is forced to peer directly with the ISP.
If you have 1.2Gbps total bandwidth and you can split it as you wish, which do you think looks better to customers? 1Gbps (+200Mbps upload) or 600Mbps symmetric? Easier to sell a bigger number.
It's honestly probably that silly.
Media companies like NBC/Sky/Universal/Xfinity/Comcast do not want you to be creators. They want you to consume consume consume... and it doesnt count against your bandwidth if you consume from THEM.
I would guess that unlike fiber -- rarely saturated in a consumer context -- people have ~always wanted more than what cable can provide and thus the operators needed to be strategic about the allocation of bandwidth between DL and UL, hence the asymmetry.
The difference is that cellular licenses generally offered equal bandwidth to up and down (ex. 10MHz down and 10MHz up). That's because they were originally designed for somewhat symmetrical communication: phone conversations where you assume each party is talking (sending) and listening (receiving) reasonably equally. With DOCSIS, it was designed assuming (correctly) that there'd be a lot more download traffic than upload traffic (which still holds true even in the era of remote work).
Upstream has always been challenging on coax. The first cable modems didn't even use the coax for upstream, instead it used a separate dial up modem.
I'm not entirely sure of the details but even after that DOCSIS has always had to use much lower frequency channels for upstream which are much more limited vs downstream which tends to use higher frequencies. Potentially because it was hard to get high frequency transmitting upstream low cost CPE devices back in the day?
DOCSIS 4 mostly removes that limitation and allows for the full amount of spectrum to be used for up or down.
A bigger issue is how many areas are so poorly served that in the present time there is no point calling it high speed.
That said ISPs tend to scale up with down so even if you might be reasonably served with 100Mbps down if you want a reasonable up you will have to upgrade to the higher down. If you intend to work from home this doesn't seem altogether unreasonable. When working for an ISP I was always surprised about how many complainers whose remote work was "so" important paid for bad speed, delivered via bad hardware, connected to bad wifi, on another floor.
> Even with good compression, video content still requires a significant amount of bandwidth. For example, YouTube recommends sending 34 Mbps for 4Kp30, and 50 Mbps for 4Kp60.
> In addition to needing high internet speed for 4k streaming, you need to factor in more bandwidth for your audio (typically in the 128–320 Kbps range) and a small amount more for overhead. Then double it — you should plan to always stay at 50% or less of available bandwidth in case of disruptions
> That said ISPs tend to scale up with down
Not in my experience, on both cable and phone lines.
For instance here is a good example of Comcasts scale in upload
https://www.connectcalifornia.com/internet-service/xfinity-p...
Astound is the same
Spectrum
https://www.reddit.com/r/Spectrum/comments/z1f5tm/dumb_quest...
Please forgive the annoying second party sources most providers have this annoying flow where you have to almost sign up for internet to see specs and such.
However, even with work from home, most users aren't pushing that much data. Teleconferencing is usually many streams in, one stream out. People working with large data feeds typically get a workstation somewhere with good connectivity and remote into it, where you send keystrokes and get big pictures. I have a pretty sad upstream, and it's not been a major problem, even if when I am pushing things from my home. Although there are a few tasks I avoided from home when I had a well connected office a few days a week.
Illegal? Under what delusional definition of 'illegal' would you classify "doing what works for pretty much everyone but you"?
I used to work with cable modems, and it's miraculous how they got this huge amount of performance out of their physical plant by basically hacking on their stuff until it worked. The cable industry has this thin layer of really smart guys and a huge number of porn hound meatheads.
I mean, did you know cable internet is basically token ring? Seriously. Not sure if that's for all implementations or just Comcast.
These wires were never meant to do anything close to this...but here we are.
From my experience this is absolutely not true. Comcast had zero issues intentionally crippling my internet connectivity because the modem I owned only supported DOCSIS 2.
I received letters for months that I needed to upgrade my modem to support a “the best speeds”. Then I didn’t, and my speed dropped to less than a megabit per second.
I could have waived it off as a coincidence, except a couple of months later I was telling the story at a happy hour and it turns out the exact same thing happened to another guy right around the same time.
Comcast does not care if you have a shitty experience. They don’t care if you call. Me not upgrading and calling support about it is just the cost of them doing business.
Sometimes installers fuck up. But Comcast (and every cable company) endeavors to not brick all of their customers in a region.
When I first got cable decades ago, it was symmetrical by default.
But they figured out business customers need symmetrical and are price insensitive, while residential customers really just need consumption and are very price sensitive.
The idea is that they can charge businesses more, and individuals less and get away with it.
This is just like airfares, where business customers travel during the week and want to be home for the weekend, and individuals will stay the weekend and are price sensitive.
ISPs like Xfinity are splitting nodes so that you aren't sharing the link with many people before you're on fiber and oversell ratios are pretty well understood at this point (and apply to all ISPs whether the wire coming to your house is shared or it just becomes shared farther upstream).
A municipal broadband network in my area has a 2Gbps connection to the wider internet and offers 1Gbps service to people's homes/businesses and that 2Gbps serves several hundreds. Clearly, not everyone could get gigabit service off that at the same time, but people's traffic is typically bursty.
The internet itself is shared at different points even if the wire to your house isn't shared. If you have dedicated fiber running from the central office to your house and then the central office only has X Gbps to the rest of the internet, you're still on a shared link regardless of how dedicated things are between you and the central office.
One huge caveat to this is that many ISPs do not mention upload speed.
I can't remember the last time it went out (outside of power outages). At least 10 years.
About 6 years ago I had some spottiness when I had moved, but a call had a senior tech rolled out, it was identified as too strong a signal as I was so close to the head-end in the building, put an attenuator on the connection to get it smack in the middle of its target range and it was perfect after that.
Even right now my bill has dropped from $90 to $80, and again recently from $80 to $75 with no contract, for 1.2Gbps down, 35Mbps up. In which I actually get the full 120% over-provisioned speed all the time. No bandwidth cap in Massachusetts, either. I run my own DNS servers in my homelab, and with the network equipment on UPS backup, even power outages have failed to cause a service interruption in the 4 years at my current location.
My area is going to soon start piloting its municipal fiber rollout soon. That I'm looking forward to as it'll give another option and increase competition. I look forward to having improved options -- likely at lower prices still, since the cities are targeting 1Gbps symmetrical for $50/mo -- one way or the other in the coming year or so.
Switched to fios and I'm never going back to coax.
Fiber cable is almost free, bundles occupy almost nothing. The electronics are not that expensive anymore as they are mass produced.
Fiber optics fusion splicers used to cost tens of thousands of dollars. Today they are under a thousand of inflated dollars.
I switched to fiber at my current home the moment it became available, but until then I enjoyed being fairly well-served by gigabit (down) cable over DOCSIS 3.1 on the infrastructure that was ready to use on move-in day.
This was a public-private partnership (funded by government but not rolled-out nor owned by our government).
The existing cable in my city is now unused (not sure how their incentives were structured, but I couldn't sign up to cable 4 years ago even though house was already wired for it)
I originally thought it was a dumb idea - state funding for netflix - however covid showed the value of the fibre connections.
https://sp.chorus.co.nz/stories/ultra-fast-broadband-ufb-ufb...
Uploading to the cloud at gig+ is also great
I do think we'll look back in a few years and wonder how we got by with just 1 gig connections. Just saying today five gig home internet feels extraordinary extreme.
I won’t miss them.
Jokes aside, I'm curious how this is even possible over decades-old cable. I get there's a new DOCSIS standard, but I'm less interested in the protocol and more interested in the simple physics of it. How can a simple coaxial cable cram so much bandwidth?
Source: happy ex-Comcast customer
A large amount of spectrum to work with and a high spectral efficiency. Wikipedia lists DOCSIS 4.0 as having 1.8GHz of bandwidth, and DOCSIS 3.1 as ~10bits/Hz. Assuming DOCSIS 4.0 is as least as efficient, thats about 18Gbps.
I haven’t actually checked to see if the limiting factor is my own network hardware but it very well may be.
Reliable internet for most is based on the quality of home WiFi network, not bandwidth or theoretical backend network that Comcast has.
Advertising board also agrees.
Apparently we can keep putting better and better signal processing on the 2 ends to interpret higher and higher bandwidth signals.
Charter Spectrum says:
Together, about 85% of Charter's footprint will be capable of offering up to 5 Gbit/s by the end of 2025 and the ability to offer up to 10 Gbit/s in some of its footprint.
Seems like the network designers want the populace to be consumers rather than creators. :(
And meanwhile here in the US, Comcast will stubbornly be attached to their coax tables until the end of times.