Hacker News Reader

In my experience, the bulk of the awesomeness of gigabit Internet is that it turns the Internet into a LAN. I think anybody who uses the Internet in a LAN-like manner can definitely benefit from gigabit speeds (and can I get 10Gbps soon, please? I'm ready!)

For instance, I (regularly) upload my 20~60GB VMWare VM files between my house and the office, because I don't want to carry a computer/disk around with me. At work, I tend to have a bunch of Remote Desktop connection to my home machines, which I leave up 24/7, as its not enough bandwidth to make a perceptible difference. And while most backup services like CrashPlan or Mozy or whatever definitely are the bottleneck, if you are backing up to your own remote servers, or to AWS via something like Arq, the speed really makes a difference, just as much as with a LAN (at least up to 100Mbps with AWS, anyway... at that point I think they become the gating factor).

I agree you don't get a huge benefit when surfing normal Internet websites, and that the benefits break down when the remote end of the connection is international, but for stuff like the above, the faster the better.

At this point in history, would users even be able to benefit from 28.8kbps connection speeds?

I have a 14.4, and I feel like the BBS's and AOL chat rooms I frequent struggle to keep up. What could I possibly need double the speed for??

Not saying 28.8k isn't great, but the internet just doesn't need speeds like that yet. Better to focus on getting everyone up to 14.4k first.

Of course we can and will use more.

But in any optimization scenario, you eliminate bottlenecks first. I question whether the ISP to home link is the bottleneck in the modern internet experience the way it was in 1998.

But that's a chicken and egg problem. Upgrading ISP to home link creates incentives for others in the chain to upgrade their link.

You're right about that, but I think it isn't purely an optimization scenario. Altogether new use cases emerge when your workplace and home are linked with such dramatically higher speeds.

Even more new scenarios shake out when work, home, and most of your friends' and family members' homes are so linked.

When connections get orders of magnitude faster, it becomes about doing new things, in addition to doing the old things faster.

I surf the web the same way with 1Gbps fiber in Tokyo as I did a few years ago with 3Mbps DSL in Ithaca, NY. But surfing the web is just a small part of what I do via the Internet nowdays.

Netflix has argued that it is.

The companies that are trying to do remote-gaming are probably having issues with bandwidth.

The days it took to upload my music collection to Google Music all argue it is.

Bandwidth isn't the only bottleneck, but it's holding some technology back.

Well, there is a difference. A 28.8 and even a 56k connection were easy to saturate with something a simple as listening to streaming radio.

I don't believe I've even approached saturating 20Mbps with multiple users in the house streaming multiple video streams and downloading as much data as they could. The other end and the in-between-bits has always been a limiting factor and has been once you get over about T-1 speeds.

Put another way, buffering in software for 28.8 was usually needed because your end couldn't stream something like a 3 minute song smoothly. Buffering on a 28Mbps connection is because the provider can't stream something like a 3 minute hi-def music video smoothly.

In principle I'd love gigabit Internet, but I honestly wouldn't know what to do with it.

I can saturate a 100mbps connection by booting a diskless computer. Why shouldn't our operating systems/environments be stored "in the cloud"? Why shouldn't I have more than one computer in my house?

Just because you can't think of things doesn't mean they don't exist :-) And once they do exist they'll seem obvious to you.

Remind me again why I can't have cameras all over my house and see them all at once while somewhere else. Why shouldn't I be able to speak to an SO wondering around the house doing chores while I'm in a hotel room?

All that aside, think of the games and entertainment that can be enabled. TVs are large enough to be almost life size now. Why shouldn't I be able to stand in front of one with a doctor or Amazon employee on the other side being able to examine each other, products, packaging etc in high resolution detail. (Note 1080 is not even close to high resolution for life size.)

Dream big.

These are mostly softballs though...1-at-a-time:

"I can saturate a 100mbps connection by booting a diskless computer. Why shouldn't our operating systems/environments be stored "in the cloud"?"

Clouds go down, your computer goes down. There's no compelling reason to put the OS "in the cloud". Your data? Maybe, if you can handle not having access to it every second of the day. And even today, one can RDP or VNC perfectly comfortably on a 20Mbps connection provided all the "cloudy" parts of the internet behave. More importantly, if the server and the links between you and the server aren't capable of giving you 100mbps then you can't saturate the connection.

"Why shouldn't I have more than one computer in my house?"

How does not having a 1Gbps internet connection prevent this?

"Remind me again why I can't have cameras all over my house and see them all at once while somewhere else."

Why can't you do that today? Streaming 1080p from x-number of cameras, multiplexed into a single display that's probably no better than 1080p is silly. The proper approach, regardless of bandwidth, is to downsample the displays before multiplexing at the broadcast side. Then you're just streaming a single video stream anyways, which is perfectly doable at 20Mbps.

"Why shouldn't I be able to speak to an SO wondering around the house doing chores while I'm in a hotel room?"

Again, why can't you do that today? There's no reason to stream video from empty parts of the house. In your use-case, you're only sending one or two video streams at a time anyway...again still healthily within the realm of 20Mpbs.

"All that aside, think of the games and entertainment that can be enabled. TVs are large enough to be almost life size now. Why shouldn't I be able to stand in front of one with a doctor or Amazon employee on the other side being able to examine each other, products, packaging etc in high resolution detail. (Note 1080 is not even close to high resolution for life size.)"

This is the only compelling case you've provided I think, but bandwidth is not the principle problem in this use-case. Everything from cameras to encoders to compressors to display devices are a bigger problem in this case.

And yes, you'll have to compress even at 1gbps. A 1920x1080 display with 24bit color at 60fps is something like 3GigaBytes per second or around 24gbps.

The next gen of TV is likely to be some variant of super hi-vision which is 7680x4320. @24bit and 60fps that's just shy of 48Gigabytes per second uncompressed or 384gbps.

So yeah, bandwidth will be a bottleneck in those applications, but 1gbps is not even close to handling it and I'm guessing the hardware required to real-time encode and compress video for a stream like that is astronomical.

> Clouds go down, your computer goes down

So what? How much useful stuff do you think I can actually do at the computer without a net connection? (Software developers are a little different, but even then there have been projects where a net connection is required to be at all productive.)

Note that it would not be law that your OS has to be remote. Not everyone would do it. But it isn't even an option today.

> There's no compelling reason to put the OS "in the cloud"

That it becomes someone else's problem to maintain is a nice one. So that it is available to me no matter where I go is nice. So that it can be demand loaded as I use bits instead of going through a multi-hour install that installs everything despite me not using a lot of it.

> .. one can RDP or VNC perfectly comfortably ..

Only if the latency is low, and you don't do 3D graphics or anything involving large screen updates. When the process and graphics card are local, they can provide superb interactive response. (I have a long history with both protocols, and even hardware acceleration for RDP - still poor performance.)

> How does not having a 1Gbps internet connection prevent this?

I meant having more than one computer demand downloading its OS. And by computer I mean almost everything with a microprocessor.

> .. multiplexed into a single display .. downsample .. reason to stream video from empty parts of the house ..

I didn't ask "how can I compromise to fit things into little bandwidth". If I had a choice between one or two cameras at a time and low quality versus all the cameras and very high quality then I'll pick the latter.

At one time 640k was enough for everyone. You could shoe horn everything most people did into that. But when given the option of more displays, more fidelity, quicker interactive response, I'll take them every time over compromises.

> Everything from cameras to encoders to compressors to display devices are a bigger problem in this case

That is all part of the chicken and egg problem. If the bandwidth isn't widely available then there won't be that much demand. There are other encoding solutions other than full "frames" - eg deltas only for random parts of the frame updated on demand at differing rates rather than having one frame rate for the whole frame. Eye tracking can determine the area to pay most attention to.

Software developers are a little different

I would say not very, while one can code without the net. The lack of libraries, documentation and other resources would make most projects a pain, if said lack extended past an hour or so. Further a lot of the software being developed, itself relies on connectivity. Now there are special focuses like embedded where it would not be as much of a pain to loose connectivity, but for 70% (I made that statistic up) of what I do as a software developer, I would probably just walk away from the machine until it regained connectivity, depending on whether I was deep in a code groove or gluing stuff together, if it's gluing then, the machine is useless without a connection.

The point isn't that Sonic is running gigabit networking to people's houses. The point is that it CAN BE DONE and on a reasonably profitable basis (I'm assuming that Sonic's management is fine with the cost outlays). Verizon was 80% of the way there and its management got cold feet.

Memo to Frontier Communications: THIS is how you do it. You can beat Verizon at their former game, get Sonic.net to show you how. (Or, even better, get Sonic.net to take over your FiOS operations in the Pacific Northwest...)

Keep in mind one of the reasons it's profitable is because running above ground wires is much cheaper than in the ground or whatever has to be dealt with in big cities. The article even mentions expanding to other towns will mean going into debt.

When their break even point is well below 2 years going into debt is hardly a bad idea. In their situation debit is simply leverage and while it increases risk you can easily maintain profitability while borrowing significant amounts of money as long as your creating capital with that debt.

The article implies that laying cable in the ground is significantly more expensive than running it above ground. Hence the need to go into debt. The ROI timeframe will also likely change with the increased costs.

The point I was making is that one of the most expensive parts of getting fiber to the home is laying the cable. In this case they were able to do a 'test' run to a neighborhood where they could just relatively cheaply run the cable above ground. I hope that their cost structure still works when moving towards more expensive rollouts, but given how we have seen other fiber rollouts get scaled back or fail I'm not so sure.

> Verizon was 80% of the way there and its management got cold feet.

I have a simple explanation for that: A regulatory and commercial environment that was clearly not going their way. Verizon spends something like $1,000 to install FiOS at your house, then a few hundred dollars a year in hardware, bandwidth, etc. To make that work for them economically, they need customers paying an average of $100+/month and preferably even more over time.

But how does Verizon get your $100+ when you watch all your TV on NetFlix, Hulu or iTunes? That's pretty obviously the way things are heading, though it may take another decade. Now all you want from Verizon is an IP pipe, and good luck convincing customers that "Internet only" should cost as much as "Internet + TV + Phone". And an Obama-appointed FCC is unlikely to allow Verizon to demand a cut of NetFlix or Apple's revenue for use of "full bandwidth": That's what the Net Neutrality fight was always about.

Without their own services to sell, the capital cost of laying and maintaining a "dumb pipe" didn't look so good.

I think Apple is going to have a go at solving this problem for the providers: "Let us take over the user experience (i.e., the hardware and software) and in return we'll give you a new way to package your services that guarantees a healthy revenue stream regardless of how consumers are getting their content." Just think how much more money AT&T makes from the average iPhone subscriber vs. the average dumb phone subscriber and you'll have the right idea.

Here's the thing I find most confusing, and it somewhat goes with what you wrote: Frontier took over Verizon's operations in Eastern Washington, Oregon, and Fort Wayne (Indiana). For a year or more, they were ambivalent or even hostile about the idea of selling FiOS TV. Now, all of a sudden, they've renewed or let stand existing TV franchise agreements and are full-out marketing it again.

I think Frontier realized what you just wrote: Being a "dumb pipe" provider--especially when your company took on a pile of debt to "buy" both aging copper and new-fangled fiber--is not a strategy for success. So now, they figure they've got the TV equipment, so why not go for it?

Personally, and this is apparently me having a minority opinion, I utterly despise Hulu, NetFlix, et al. Why? None of them has matched the ease of use of my TiVo connected to a linear cable lineup. Television is supposed to be an "idiot box" and for me and the other two people in my household, TiVo + FiOS TV = astoundingly easy.

My guess is that any full-fledged "Apple TV" will take your preferences into account. Here's my wild theory:

How many unique TV channels are there in America? OTA and cable. Maybe 2,000? With a ton of redundancy in the programming. What if Apple simply recorded all of them into that big fancy data center? No need for a DVR: Now you can stream anything your cable provider offers any time after it airs whether you have a "season pass" or not.

There will probably be a few annoying restrictions (Shows disappear after two weeks?) and you'll probably have to watch ads. But as far as I can guess, it's the only way Apple can give you time-shifted access to non-streaming programming (e.g., the NFL) without reinventing TiVo, which I'm pretty sure they don't want to do.

It all hinges on whether they can get the contracts ironed out. But the cable/fiber companies will see dollar signs from more expensive "Apple TV" plans (with two year contracts to make the set cheaper?) and the networks may be swayed by the promise that customers will have to watch ads, or pay them to make the ads go away.

Current speeds are not really enough for HD video. Even if you get 1080p video online it is still highly compressed for online transmission. If I remember correctly I think blue ray is 40-60 Mbps. Higher compression means there simply isn't as much detail.

Over the air HD in 1080i format is generally about 20Mbps. It looks pretty darn good to most people.

Yes, higher bitrate makes better pictures, but for TV, over the air is currently the best way to get network programming in the least compressed format.

Build it, and they will come. Expanding this kind of service to the majority of domestic internet users would be a game-changer for the types of new apps and sites it could enable.

I recently moved from Chicago to New York, getting 6Mbps DSL at home in Chicago and 1Gbps (yes) Internet at work. The difference is noticeable.

Regardless of need, once you have fiber to your building, 1Gbps and 20Mbps cost the same. It's getting something other than a POTS circuit that makes the difference.

That's the genius of Sonic's plan. Advertise gigabit, customers use 20 Mbps, everybody's happy.

Why does that make everybody happy? What I got from the article was that the difference in cost to the ISP for having someone connected at all and using a ton of data had grown increasingly minimal in recent times.

and that the cost difference from somebody at 20mbps to 1gbps is nominal. It does not cost them that much more in true usage. I think it is pretty straight up of them to have their installers disclaim to customers that they really don't need a 1gbps connection.