Will Netflix Destroy the Internet? 577
nicholasjay writes "Netflix is swallowing America's bandwidth and it probably won't be long before it comes for the rest of the world. That's one of the headlines from Sandvine's Fall 2010 Global Internet Phenomena Report, an exhaustive look at what people around the world are doing with their Internet lines. According to Sandvine, Netflix accounts for 20 percent of downstream Internet traffic during peak home Internet usage hours in North America. That's an amazing share — it beats that of YouTube, iTunes, Hulu, and, perhaps most tellingly, the peer-to-peer file-sharing protocol BitTorrent."
So, how long before... (Score:5, Interesting)
Re:Will posting inflammatory headlines (Score:3, Interesting)
And yet it gets tons of page views. The bottom line is that the parent company has chosen to go more after dollars than making a niche group happy. Take from that what you will.
What happened to the Dark Fiber? (Score:3, Interesting)
Re:So, how long before... (Score:3, Interesting)
netflix will price itself out before it happens (Score:3, Interesting)
i have netflix and the streaming selection is pretty bad compared to the DVD selection. the reason is that they haven't struck deals with most content creators yet.
my cable bill is $130 a month for TV/DVR/Internet/phone and from what i've read approximately $30 of that goes to the content creators. for netflix to offer all the content there is they will probably have to raise their prices as they strike new deals for more content, especially if it will include movies and new TV shows that just played the night before.
if i wanted to dump cable i'd have to pay more for a la carte internet and more to AT&T to increase my cell phone plan to unlimited minutes. it would kill the entire deal since it makes more sense to just pay $10 a month for a DVR
and this theory is based on just he financials of striking content deals. netflix will have to pay a lot more in bandwidth costs as the amount of content increases.
i don't understand the entire streaming fad. it's only around because the cable companies are always a few years behind. with digital/HD cable what you watch on your cable box is essentially streaming except it's a lot more efficient than netflix's TCP/IP over the internet version. the cable companies just need to update their software and service selection
The answer is - Never (Score:5, Interesting)
Never.
Netflix is not Bittorent and has a well defined source which is a commercial entity. So the ISP knows after who it needs to go. Further to this, as it is not P2P traffic Netflix itself has no choice but to grow its infrastructure if it is to retain its service level. Otherwise it will congest its links to ISPs and kill its own service offering.
So Netflix will have to start building its network infrastructure and peer with ISPs close to the user across the US and the globe.
We have already been through this. Before it was Google/Youtube destroying the Internet. Well it did not. Simply Google now has a backbone which can put most tier 1s to shame and peers with anyone anywhere.
Most importantly, the number of links and peerings will increase so the end result will be GOOD for the Internet as it will become more resilient (Assuming ISPs use local/distributed peering not just for Netflix but for the other peering).
Re:What happened to the Dark Fiber? (Score:4, Interesting)
it was all bought up long ago and there was an article here a few weeks ago how most of it has been lit up and the bandwidth has almost been used up.
Re:What happened to the Dark Fiber? (Score:1, Interesting)
The problem you have is Last Mile Infrastructure. The backbones of top level ISPs can handle this stuff without even breaking a sweat. But the ISPs that residential customers deal with, Comcast et. al., get rather stingy when it comes to making sure their last mile infrastructure can even handle a fraction of the numbers that their marketing departments spout all over the place.
Re:I can confirm this story (Score:3, Interesting)
Unfortunately that something is going to target the consumer because the government lacks the balls to tell a corporation to go fuck itself and compete for business.
Re:What happened to the Dark Fiber? (Score:3, Interesting)
His point is that need is increasing.
Dark fiber is dark because in areas that needed 1 fiber, the additional cost to run a bundle of fibers was miniscule. (Labor costs to lay the fiber dominated the material costs.)
Probably on the line of 1 fiber might cost 100 million, and 100 fibers might cost 101 million.
So if we go past the capacity of one fiber, we can in theory light up another. In practice, there might be space constraints at the endpoints that weren't thought of when the big bundles were laid down.
Re:So, how long before... (Score:5, Interesting)
I don't know. This could be a GOOD thing. Previously, there seemed to be some stigma attached to high bandwidth users. Anyone who was using a lot of bandwidth was "obviously" doing SOMETHING shady. With the birth of services like this, it's starting to become quite common for regular old users to suck-up lots of bandwidth. I think the ISP's may finally have to pony up some dough and upgrade their infrastructure.
Of course, if they'd had a bit of sense, they'd have realized a simple truth that applies to almost any computer usage, be it processing power, bandwidth, or anything else: today's power users use what tommorow's regular users are. Rather than trying to persecute your heavy users, use them as a metric to gauge what you need to roll out.
I think he means things like cache engines (Score:5, Interesting)
You go to the big (and maybe even small) ISPs and say "We'll provide you with hardware to store Netflix movies. When customers request movies that are on there, it'll come from those, rather than our servers. We pay all the hardware costs, you save on bandwidth."
Akamai does just this. They peer with all sorts of people to get their cache engines in ISPs. At the university I work at, they came to us. The deal was they'd provide the computers (3 servers last I checked) and a switch. We set up our networking to go to those first. Net effect is when you ask for something that has been cached on there, you get it locally, rather than from one of their server farms. Keeps their bandwidth costs down, our bandwidth costs down, and increases speed. Now not everything is stored there, they host a lot of shit. I don't know how their computers decide what to keep where. Some popular things (like Microsoft updates) I think get auto cached, others I think it is based on demand. However even with just a fraction of their content cached, it makes a big difference in bandwidth.
Netflix may need to start doing the same. I mean video is the ultimate in things that could be multi-cast, except that we want it on demand. Well cache engines work well for that. Since the video never changes or gets updated you push it out when you get it, and then those serve it up to people as often as they want it.
Comment removed (Score:5, Interesting)
Will Ford Destroy the highway system? (Score:5, Interesting)
Your intrepid reported, reporting from 1910...
Many people are reporting the growing difficulty of navigating their horses and buggies through the town streets due to the growing presence of noisy and fast moving motor cars made by Henry Ford. Predictions are that because of this obnoxious growth in motor cards that our highways will become completely unusable within 10 years!
Re:So, how long before... (Score:5, Interesting)
I think all ISPs realize that bandwidth needs to increase. Looking at a bandwidth graph over the last three decades would make that plainly obvious.
However, how will this be paid for? They say it should be the Googles and the Netflixes, I say it should come out of their CEO's new yacht fund. That, I think, is where the point of contention lies.
Re:I think he means things like cache engines (Score:4, Interesting)
But one of the biggest ISPs is Time Warner Cable. They are certainly **not** going to help
Netflix deliver unlimited movies for $13/mo, when TW charges a lot more for the same thing.
Re:So, how long before... (Score:3, Interesting)
No. 1080p with AC3 surround compressed at a mpeg4 is only 12GB per movie. Split out the AC3 into separate tracks and compress the crap out of it to 160kbps and you can lower it by another .5GB. and this low bitrate is acceptable to 99% of all netflix users.
But, here is the fun.... you get 720p max from netflix... so it's actually 5-6Gb from them.
Have you actually tried to do any bluray compression to mpeg4? 25GB is for uncompressed or mildly compressed only 2 Blurays I have ripped have had any main feature above the 25GB mark. Most blurays are compressed to hell.. AVCHD is the darling for bluray.
It's why I always though that HD-DVD was better. simple MPEG2 on the disc delivered far better picture with less hardware power.
Re:So, how long before... (Score:2, Interesting)
The old cable-TV model is slowly collapsing for a few reasons:
The basic architecture of one pipe shared by whole neighborhoods is inherently bandwidth limited and not scalable.
In the sub-nets where the Internet signal is sent over coax along with TV signals (not the fiber backbones), the interference (intermodulation distortion) resulting from large numbers of signals originating from the customer’s modems reduces bandwidth quickly. Cable is inherently one-way, and does poorly when pressed into bi-directional service.
New Internet companies are able to distribute media ala-carte at much lower cost. Partly because they don’t have the contractual obligations to distribute content. The dispute between Fox and Cablevision is but one example of the greedy content providers forcing all cable customers to pay, whether they watch the content or not.
Demand and use of high speed Internet and high resolution HD channels is increasing rapidly.
Services like Verizon FIOS have a major edge over the antique cable system as they have individual pipes to each home and can increase total bandwidth with less infrastructure.
Re:So, how long before... (Score:3, Interesting)
I thought networking technology also evolved. I mean 15 years ago I had a 166 pentium MMX and now I have a 3GHz Core i7. 15 years ago my LAN was 10mbit on Coax but now I have Gigabit.
But for some reason, ISP gear doesn't seem to grow as fast as consumer stuff? Cause I keep hearing about T3s and OC-3s and 622mbps. How much fibre capacity unused out there,d ont you think? I mean, a 12-core fibre carries 144 strands, or 72 full-duplex connections. Almost 10 years ago, a Cisco 12000 series router could push 40gbps in each of those. 6 years ago, Cisco introduced the CRS-1 router which could switch 92Tbps, and they have the CRS-3 now (last time I checked).
I thought the largest carriers didn't even PAY for bandwidth, as the traffic was kept inside their own network most of the time? And when it reached another network, they usually had a peering agreement for a VERY low price?
Now I'm comfused
(Sarcasm)
Re:So, how long before... (Score:2, Interesting)