Netflix CEO: In Ten Years, “We Will All Have A Gigabit To The Home” | Hacker News Reader

Netflix CEO: In Ten Years, “We Will All Have A Gigabit To The Home”

techcrunch.com

50 pointscosgroveb78 comments

"All" is just a wee bit hyperbolic.

I live in a rural county outside of Lawrence, KS. The fastest 'net connection I can get is Verizon 3G. In a decade, I'll be lucky if WiMax has made its way out that far, but there's absolutely no reason why a telco would run a physical pipe out to me. It'd be miles of cable/fiber with me and my chickens on the other end of it.

People who live in cities and tech hubs forget the fact that there's a lot of open space in this country. Much of the population may very well might have gigabit in a decade, but there'll always be a portion of the population who just live too far out in the sticks.

Came here to post a similar story. 26 miles from the white house here, but in a lower density area. My home internet comes from a T1 that runs about $500/mo to provide 1.5Mbit. My other options are ISDN (128kbit) or sprint EVDO at about -100db, providing ~200-500kbit but with many, many dropouts. 5 years ago there were the same options. 10 years ago it was T1 or ISDN. 15 years ago the same.

Nobody wants to serve below a certain density with anything.

Comcast was asked by the county to provide a customer pays dig fees estimate. They wanted 2.2M up front. Verizon actually wants to stop maintaining the copper plant in the next few years. It's very possible that this would leave us without enough clean pairs in the neighborhood to use leased lines. They indicate that we're scheduled for FIOS somewhere around 2020.

Verizon actually wants to stop maintaining the copper plant in the next few years.

Verizon wants to do this everywhere, not just in low-density area's.

You may be technically correct, but from the perspective of Netflix, it's economically "all" in that the addressable market of US households without Gigabit ethernet will be so infinitesimally small to not be worth worrying about when thinking about customer acquisition/retention or expenses.

I'm surprised that you don't have access to dsl but do have access to 3G, I live in a pretty rural part of my state but dsl is still available (but my first real hop is 80-100ms) but it is 1mbit. My only other option is satellite internet or purchasing a t1 for ~300/month.

I must say though if the cell networks stop charging so much for access I could see a real use for it as it is much faster than satellite internet and if they encompass the whole country in 3G broadband for all would be more of a possibility.

> I'm surprised that you don't have access to dsl but do have access to 3G.

There are quite a few locations (including mine) where the remote infrastructure is fractured (for lack of a better term). The remote box out here, a Lucent Mod 5, is quite capable of housing a DSLAM, and I am certainly within a functional distance from the Lucent box (4k-5k feet). The problem is the distance from the Lucent box back to the CO (6 miles) which is on copper T1 circuits. No fiber out here. The ILEC, AT&T Florida, appears to have no interest in replacing the T1 cable with fiber, nor to populate the box with a DSLAM.

There is an ISP that has deployed 2.5GHz WiMax, but I'm on the fringe of the range, plus I have vegetation issues.

When I see a title like the one at the top, I always cringe, because I know full well that saying 'everyone' will get this is hyperbole.

FSO?

http://en.wikipedia.org/wiki/Free-space_optical_communicatio...

Did the local electric utility decide it wasn't worth running 120V power to your home either? I don't know of any reason that running fiber is any more expensive than running power lines (if anything, being able to piggyback the fiber on existing poles should make it cheaper).

They likely did. Electricity was pretty much nonexistent in rural areas until the 1930s when the Rural Electrification Administration provided loans to cooperatives to bring electric utility service to areas too sparsely populated to justify the costs for a traditional utility.

Yup, nailed it. Electricity didn't come to my area until the late 1940s; as far as I can tell, my house wasn't actually wired until the mid-60s.

I lived in Manhattan, one block away from central park. My only reasonable internet option was Roadrunner 10 Mbit. No FiOS. I seriously doubt that gigabit claim.

What was your option 1 decade ago? I bet it wasn't even 1 MBit.

This doesn't surprise me, I live in San Francisco, and my neighborhood has only one option for internet connectivity: comcast cable. Well, that or tethering via my cell phone.

And Comcast in San Francisco (at least in the mission) offers a 50mbit connect that (when I've had it in the past) works better than advertised. I think peak downloads we saw were just over 7MB/s.

What's the problem?

That Comcast is your only choice, and if you disagree with their business practices, usage caps, pricing, or 'network management' that the alternative is dial-up?

That's like arguing that electricity isn't available because you don't agree with how PG&E operates. Just because you don't want to give them your money doesn't mean they cease to exist.

The point remains that Comcast is still offering connections that are quite fast, and it's completely feasible (and expected) that they'd reach gigabit connects in the next 10 years.

Have you checked DSL options like AT&T or Speakeasy? I would be very surprised if that wasn't available in your area given that it goes over phone lines.

Sonic.net Fusion is the way to go for DSL in SF.

How odd to see a prediction like this on the same day that AT&T announced new bandwidth caps for its residential broadband services. Which some industry watchers speculate is partly intended precisely to impair streaming video services. Like Netflix.

http://money.cnn.com/2011/05/03/technology/att_broadband_cap...

I suppose it's possible for you to have a 1 gigabit connection, but you're entitled to only 2 gigabytes per month...

So you can use your entire capacity in 16 seconds. That would be bad.

Even AT&T can't stop inevitability, as much as they will stomp and pout.

Supersonic jets flying routes all over the US were once considered inevitable. There was once a time that skyscraper height increases were considered inevitable. Bigger ones were put up all the time. Audio quality on phones was once an area where improvements were constant. Now I'm surprised if people even mention call quality in phone reviews, except to say that it dropped fewer calls.

My point? What seems inevitable today can often seem like a concern of a bygone era.

The points you make are economic in nature. Supersonic flight is just too uneconomical to really be worth it at this point in time---regular jet service is "good enough" (just a side note---some thirty miles west of Miami a two-mile long runway was built for supersonic flights; it's now used for training. You can see it here: http://maps.google.com/?ie=UTF8&ll=25.861695,-80.871048&...)

A friend of mine, an architect, says that it's technically feasible to build a skyscraper a mile high (not to say it wouldn't be technically challenging; just that it is possible). But they aren't not because it's not worth the money to build (hard to rend out the office/hotel/living space where there aren't windows).

And for phone quality, mobility and convenience of use has trumped quality for most people, so again, it's economic in nature.

The points you make are economic in nature.

Indeed. And so is bandwidth to the home.

But there's a killer app for fast bandwidth - high definition, low latency videos. What Netflix is talking about. When consumers see that, they'll want it and they'll pay for it in a way that they would not pay for a supersonic jet flight.

The problem is that this need is already served well by cable and satellite service. If video is the only killer app you have for this bandwidth then it will never get built.

Inevitability of what? As time has passed, I think it's become less likely that we'll see any significant improvement in bandwidth availability in the US rather than more likely. ISPs are trying there hardest to implement data caps. Rollouts of fiber to the home seem to be slowing down rather than increasing.

I live in Portland, ME. Not exactly a huge city, but it has a decent population and isn't really that far from Boston. Yet I see no indication that the city will receive any sort of fiber to the home in the next decade. And if it's that bad here, I can only imagine how bad a shape more rural areas of the country can get.

Of course they can, especially in individual cities where they have an effective monopoly. For instance, my uncapped DSL in 1997 in Iowa was faster than the AT&T DSL I have in San Jose right now in 2011. Why is that? It's not technical limitations, I can tell you that.

Fast internet does very little good when we have 250 Gb/month transfer limits imposed on us by Comcast (Xfinity) and AT&T. Those companies are preventing progress with such caps.

I see a future where the country is increasingly segregated in terms of bandwidth and connectivity... we're already seeing this in some areas, but expect it to accelerate.

Large cities will be wired for gigabit connections in the home and LTE wireless while rural areas will be increasingly left behind with connections averaging in the single megabits.

As computing power and bandwidth requirements accelerate, these areas of the country will be affected disproportionately.

How is this different than any other "service"?

Its way easier to get everything from Thai food to dry cleaning to auto parts in a city.

There are certain services which are available everywhere with relatively equal quality: power, mail, water, voice cellular phone, and wired telephone.

One server that isn't available, despite subsidies given to build it out, is high speed broadband- both wired and wireless.

>Its way easier to get everything from Thai food to dry cleaning to auto parts in a city.

Actually it was easier in the remoter parts of Ohio for me to get parts than it ever has been in NYC or SF.

Maybe if Comcast goes bankrupt and gets replaced with set of small companies. Otherwise, I doubt it is in business interest for large cable and teleco companies to increase bandwidth. Profit and increasing bandwidth don't go together when you have monopoly.

Maybe they need to be busted up. I remember my local cable company before Comcast took over.

Of course, I would be happy with some muni. ftth.

I live in an area (not far from Sacramento) where half of the residents are still on dialup because there's nothing else available.

Just a data point.

If nothing else is available, what are the other half on? Nothing?

No. There is (some) DSL and Comcast availability, but no fiber anywhere. To be more specific:

"I live in an area (not far from Sacramento) where, of the residents that have internet access, roughly half are still on dialup because they can't get anything better."

And, of the dialup subscribers, many have problems every year because it's old infrastructure and the aerial lines have a tendency to fill with water in the winter.

I only point this out because I suspect that there are a lot of people living in metropolitan areas that really have no idea what internet access is like in rural areas.

He's off by an order of magnitude. It took about 10 years for my cable internet to go from 100KB/s to 1MB/s where it stands now (this is in Canada). That's a 10 fold increase in 10 years, not 100 which he is predicting. So by 2021 I expect to have 10MB/s, not 100ish. And unless some anti-UBB legislation gets passed I also expect my bandwidth cap to still be where it is today (90GB/mo).

I disagree - We already have +10Mbps offerings from Comcast etc.

The issues is moores law - we will be at multi terabyte backbone links in ten years.

The access port in most environments is already 1 gig. (Enterprise environs)

You will have a 1 gig link - but you will be metered to some <1 gig feed - and, as we do now, pay for bursting.

If you're going to use Moore's law you'd look at what your speeds were 10 years ago. On a cable modem in 2001 I was certainly getting more than 100kbps (~12KB/s), I was probably closer to 1mbps.

But thats not how moores law trickles through the tech stack to the consumer....

We get shifting increase in computing benefits all over the place -- the last place the consumer sees the benefit is in access speeds.

Carrier backbone links will continually grow, but consumer access speeds will grow at a much slower rate.

The compute power of all devices will grow incredibly fast by comparison (moores law).

The reason for this is that you already pay what the market will bear for your access. The infrastructure costs to the carriers are incredible at the access layer, so the cost benefit for them is to upgrade much slower.

I predict that the disruption will come from the following potential areas:

* Google, Facebook or Amazon will offer a fiber to the home network soon.

* Advances in Spectrum technology cost/availability will offer the Gig-link

* Municipal networks are seen as the next New Deal and massive technical infrastructure projects are sponsored by the government because they do know how to supply funding for infrastructure and dont understand how to fund startups.

* Google, Facebook or Amazon will offer a fiber to the home network soon.

This would be a nightmare working with various municipalities and their right-of-ways. No two situations would be the same. Anytime a road is repaved, the fiber has to come out and be replaced. Lots of maintenance, too.

* Advances in Spectrum technology cost/availability will offer the Gig-link

Not possible. Has nothing to do with Moore's law. Imagine a family of 5, each watching a separate live 1080p stream. Now, how many people in a town can watch a unique 1080p stream simultaneously?

Yes, this will save the United States. I'm serious. This is the only viable option.

I think that you will be surprised with respect to the fiber to the home projects in the near future.

Some newer housing developments have fiber-to-the-home, with blazingly fast internet access in perpetuity, included in the price of the home. I have no idea how the financials actually work on that. I don't remember what the speed test was, but I half-wondered if it had malfunctioned somehow because I'd never seen anything like it in a residential area.

That said, I suspect that very few people are going to get that kind of access relative to the population as a whole. The U.S. is pretty big, and there are a lot of people in areas that simply aren't feasible to do expensive build-outs. (I live in one such area.)

I seriously doubt it. Perhaps 50-100Mbit/sec which is obtainable using the cable/telco infrastructure already in place. I don't see any killer apps on the horizon that will actually require a gigabit of bandwidth into the home. Even today you can get pretty good HD video over most residential broadband connections. Even if we double or triple the resolution/bit rate the math still doesn't work out. Other than people lusting over a bigger number what's going to justify anyone spending tens of billions of dollars to make this happen in the next 10 years?

At the Verizon store today, their "Blazingly Fast" connection was 0.2mbps on speedtest. We all know technology exists to go much faster than that, but it is the politics of the companies involved that is slowing everyone down. Consider the other news story submitted to HN recently about 1Gbps coming to Chattanooga not Verizon, not Comcast, but the community-owned EPB. Whether it is here in 10 years will depend not on the tech, but on the companies so it could come sooner than 2021 if there is a shift with the sources' understanding.

Moore's law is an observation, whereas Shannon's law is basic physics. Shannon's law dictates the maximum bandwidth over a physical communication link. Telephone wires and over-the-air solutions aren't going to cut it. The only two was to get Gigabit Ethernet to the home are optical fiber and coaxial cable. Most communities won't be able to afford fiber installations, so Netflix and other streaming sources will be 100% at the mercy of Comcast and Cablevision. The future of home entertainment is the cable companies, not Netflix.

Shannon's Law doesn't actually say anything about the medium. That is we can find more bandwith OTA. I'm sure 30 years ago not many would have guessed you could do 100Mbps OTA (even knowing Shannon's Law), yet with LTE we can.

With better engineering (multiple frequencies, compensations, etc...) I have little doubt that we can achieve bandwidths that are still technologically out of our reach today.

With that said, when it does end, I'll probably be just as optimistic not knowing we hit the wall.

It's not just the speed, it is how many simultaneous connections at the speed that is important.

Also, 30 years ago, a MicroVAX was running at 5MHz, so anyone optimistic enough to assume we would break the 10MHz CPU barrier, could also assume we would break the 300bps communication barrier.

Netflix, of course, needs this (or some fraction of it) to be true. Meanwhile, all the current (lazy, inept, consumer-hostile, mediocre) bandwidth providers are going to try to fuck them out of the game.

So a big part of me wonders what Netflix is working on behind the scenes to secure their future from a bandwidth perspective. This is a company that has always been so proactive – I have a hard time believing they're twiddling their thumbs and hoping that AT&T, Comcast and other telecom companies will be menschy guys.

What could they do? I know very little about large scale networking. Is there some supply-side critical path that a company like Comcast needs that Netflix could buy for leverage? On the consumer end, is it reasonable to imagine them reselling bandwidth to end-users under terms that favor their business, thus spurring competition?

Everyone who controls the consumer side of bandwidth seems to be an asshole of one flavor or another – perhaps Netflix's needs could align with consumers in such a way as to change that.

Could they start running fiber to people's homes?

Whatever happened to Google's "experiment" of running a fat pipe to one community?

Nice, that'd be convenient! I'd only have to deal with one company to get both my internet connectivity and my tv/movie content. Wait a minute...

Google announced the winner as Kansas City last month. http://www.businessweek.com/technology/content/apr2011/tc201... has some interesting notes about the implications of Google running the service and disrupting the ISP market. One strategy they talked about was using Kansas City's own infrastructure to run fiber through utility poles and already present street conduits to get fiber to the home and cut costs.

Just over a month ago, on March 30, they picked Kansas City, Kansas.

Google Blog: http://googleblog.blogspot.com/2011/03/ultra-high-speed-broa...

HN: http://news.ycombinator.com/item?id=2387866

Google has announced they'll be offering gigabit internet in Kansas City, starting in 2012.

IIRC, a post to NANOG a while ago claimed that the cost to run fiber to the home is well over $1k/household. I think the math works out to the conclusion that neither Netflix nor Google has enough money in the bank to run fiber to a large fraction of America.

So the question becomes: if you get fiber to 1% of America, is that going to be enough of a demo project to convince someone to come up with the money for the rest of America?

There's no way Netflix could afford to do that. Most of their cash will need to go to licensing deals just to stay ahead of the competition. As streaming video becomes ubiquitous, content will be the differentiating factor.

I would sign up for a fast pipe from Netflix pretty much sight-unseen.

One of the best and cheapest things they can do is seed the expectation of lots of bandwidth in as many people as possible so that they then put pressure on the telecoms.

As has been mentioned, there are many communities who still are only served by dialup, or slow-ish cable/dsl connections.

I wonder if the telecoms are waiting to roll out fiber optics to the everybody, and potentially skipping over installing the intermediary fast dsl/cable connections for the rural communities.

Sometimes it feels like we don't even have a Gigabit to our country, even if we were allowed to use Netflix...

(New Zealand)

How many servers will it take to push everyone data to a billion people all trying to pull data at a gigabit per second?

I'm confused whether to feel elated, sad or hopeful at the 10 year timeframe.

Canadian ISPs say: You can already download 1 Gigabit. 2 Gigabits are $25.

In Australia we may have it some time in the next 30 years after our multi-hundred-billion dollar FTTP program is complete (I include modest 400% time and cost overruns in my prediction).

I disagree with your (hyperbolic?) overrun predictions, but won't international backhaul remain a problem? Then again, a more connected country might make putting proper CDN points of presence more attractive.

You're correct in highlighting backhaul; when talking to other Australians I point out that it would be useless to have fibre running into every home if we're still squeezing everything down 3 (soon 4) very expensive pipes.

Out of the 50 odd billion that's been budgeted for the project it might have been nice to splurge a few billion for an additional 2 or 3 pipes -- say one to Singapore, one to Guam and another on the trans-Tasman route to California.

The paucity of competition in backhaul makes traffic so expensive in Australia that I actually host all my sites in the US -- including several of Australia's most influential blogs. Dollar for dollar I can get the same RAM, CPU and disk space in Australia, but I will get only about a tenth of the traffic quota with punitive overages.

I do exactly the same in NZ, except here we have 1 intl. pipe. They're trying to get another one off the ground, but it's probably far off in the future and I'm yet to be convinced it will make any meaningful difference. The govt. are also trying to roll out fibre to the home, but they're fucking it up fantastically and basically gifting it to Telecom (NZ's monopoly telecommunications provider). Thank allah for Linode.

I'm on Linode too.

So if backhaul was being built, perhaps NZ could be persuaded to chip in for more trans-Tasman pipes being laid? As I understand it, the one pipe from NZ to California continues on to Sydney.

Actually, looking at their website, looks like they're basically planning on duplicating that pipe and running it to SYD as well http://pacificfibre.net/ Guess that could be good for you guys too.

The big cost for hosting providers in Australia is bandwidth. Broadly available fiber should help that.

I disagree about the need for the government to pay for international bandwidth. So far private companies have done that pretty well, and it's quite a competitive market.

More demand will probably encourage another couple to be built - but without needing tax payer money.

My point is that the entire economics of IP traffic in Australia is driven by those international pipes. Massive internal bandwidth won't change that (less the effect of CDN POPs).

It is becoming a competitive market, but so is DSL. My point is that we're going to spend tens of billions on a white elephant we might as well actually spend some of it where it counts. Even if it took the form of giving (say) a billion dollars to each of PIPE Networks, Southern Cross Networks and Telstra to get them to lay more cables.

I don't believe that there is actually a shortage of TransPac bandwidth, and the prices aren't crazy. The high prices for hosting in Australia is because of the cost of domestic bandwidth, not TransPac. (As a random example - many people host using Amazon EC2 on the US West Coast instead of in Singapore because some ISPs route to Singapore via the US for cost reasons(!!)).

I do agree more international bandwidth would be great though, but I'd prefer to see additional entrants rather than money to the existing ones.

It would be interesting to see if TransPac bandwidth usage follows a powerlaw, eg: most usage to a few sites (YouTube?)

If this was the case then it would make a lot of sense to encourage those companies to setup a local CDN endpoint, which would also reduce the usage of the TransPac links.

Domestic bandwidth is expensive because ISPs and data centres don't distinguish between domestic and international packets at the server, but get billed by the international pipes based on traffic. Hence the price for bandwidth consumers -- retail and commercial -- is set at a level that covers that international traffic.

Hence domestic prices are driven by the international component. These conditions should continue for so long as there are a) few competitors and b) most content is located overseas.

I agree that it would be nice to have local CDN POPs, but all that does is transfer the high cost of Australian packets back to the provider ... and they would want to pay that because ...?

I agree that it would be nice to have local CDN POPs, but all that does is transfer the high cost of Australian packets back to the provider ... and they would want to pay that because ...?

Faster services -> more usage -> higher revenues (cite Marissa Mayer, Google, circa 2007 about search speed and revenue).

I suspect that most large internet companies have crunched the numbers and decided that the small Australian market doesn't justify the large Australian costs.

Australia is an internet minnow.

This is what I love about Netflix, both their CEO and from Adrian Cockcroft.

They are visionaries who are really thinking in terms of 5 and ten years out.

Their business model WRT IT is fantastic, they KNOW that computing is a commodity utility that Netflix has no business being in - they are a content provider.

They are effectively and very successfully both consuming cloud computing and driving its direction.

If you look at Adrian's presentations on the subject - they clearly have a solid understanding of their technical needs to support a visionary content delivery model.

This is why Netflix will succeed and dominate in the future.

They can focus on the way that content is successfully delivered via IP -- they are already YEARS ahead of other players in this perspective, where CATV broadcasters have lots of legacy debt.

We have already seen the first small battles in the space with respect to net neutrality/QOS manipulations from carriers such as Comcast.

Further, the added benefit is that companies like Amazon benefit from the input and partnerships with Netflix allowing both offerings to co-mature through situations like last weeks outages.

What will be interesting is to see the product offerings from Netflix expand.

Wait till we hear they bought a movie studio. They are doing a show - other content creation ventures are sure to be close behind as the biggest issues they have are licensing.

When they own the license to content delivered over their amazingly efficient delivery infrastructure a new content era exists.

Wait until they offer content hosting and delivery via enterprise accounts. Imagine when they can host and deliver content for channels with a whitelabel service -- or they create a video CDN for content.

They should, if they already are not, work to foster and host indie movie/docu content and in return get a better voice on the licensing models of such content.

Reply on news.ycombinator.com