Netflix Is Eating Up More Of North America's Bandwidth Than Any Other Company
another view might be that netflix's customers are eating the bandwidth randy
On Wed, May 18, 2011 at 7:56 AM, Randy Bush <randy@psg.com> wrote:
another view might be that netflix's customers are eating the bandwidth
randy
One of the UKs large residential ISPs publishes what their customers use bandwidth for at http://www.talktalkmembers.com/content/view/154/159/ "Streaming protocols" do use up a large % there, but only 2.9% is listed as used by BBC iPlayer (like a no advertising version of Hulu, but only for one broadcaster), Rapidshare and Facebook are 1.9% each, whilst YouTube is 9.7%. It's kind of interesting.
"Eating Up" sounds so overweight and unhealthy. Since a good number of us get paid for delivering bits, isn't this a good thing? Always glad to see bits and dollars flowing into the Internet, personally. However must express severe dissatisfaction with the topic of the thread a while ago referencing Comcast trying to charge providers for delivery over their network. Maybe I'm wrong, but I'm pretty happy with the current model... even if it means a $5/month residential rate hike (or something). --C
-----Original Message----- From: Carl Rosevear [mailto:crosevear@skytap.com]
"Eating Up" sounds so overweight and unhealthy. Since a good number of us get paid for delivering bits, isn't this a good thing? Always glad to see bits and dollars flowing into the Internet, personally. However must express severe dissatisfaction with the topic of the thread a while ago referencing Comcast trying to charge providers for delivery over their network. Maybe I'm wrong, but I'm pretty happy with the current model... even if it means a $5/month residential rate hike (or something).
--C
Well it depends if Netflix pay for the bandwidth they use or if they get it all for free with non settlement peering. If, suddenly, your business model breaks because of a huge demand for high bandwidth services by your customers then either you need to charge your customers more or Netflix (or whoever) need to share the pie. -- Leigh Porter ______________________________________________________________________ This email has been scanned by the MessageLabs Email Security System. For more information please visit http://www.messagelabs.com/email ______________________________________________________________________
Leigh Porter (leigh.porter) writes:
Well it depends if Netflix pay for the bandwidth they use
You mean, customers have to pay for the bandwidth they use. I'm sure NetFlix is paying *their* network and other transit providers for outgoing bandwidth they consume.
or if they get it all for free with non settlement peering. If, suddenly, your business model breaks because of a huge demand for high bandwidth services by your customers then either you need to charge your customers more or Netflix (or whoever) need to share the pie.
Whoever ? Nah, the consumers. Bad business model, change business model. Phil
On Wed, May 18, 2011 at 12:32:49PM +0200, Phil Regnauld wrote:
Leigh Porter (leigh.porter) writes:
Well it depends if Netflix pay for the bandwidth they use
You mean, customers have to pay for the bandwidth they use. I'm sure NetFlix is paying *their* network and other transit providers for outgoing bandwidth they consume. Phil
note the classic Man-In-The-Middle attack here. Or in other words, the ITU half/circuit billing model for traditional telecomunications companies. The telecom model is : "I'll provide you with a tranist path to me, and trust me to hand your communications to the other party you wish to communicate with." So GTE / MaBell gets to bill -both- parties at their usual usarious rates. The problem here is that the incumbent operators have and are fighting tooth/nail to ensure their near monopoly on access. So... We either need to re-regulate them to assure equal access at equitable rates -or- we need to de-regulate the access market and open up last mile ROW to all comers. What we have done is de-regulate the access and retain the monopoly status on last mile ROW. the incumbents have captive markets and can charge whatever the market will bear. Great work if you can get it. If we truely beleived in end-2-end, we might see more systems using or trying to find other access paths ... YMMV of course. /bill
On 18 mei 2011, at 12:06, Leigh Porter wrote:
Well it depends if Netflix pay for the bandwidth they use or if they get it all for free with non settlement peering.
The whole point of peering is that both sides benefit. A bit like one bringing the traffic to the half way point and the other taking it the other half of the way. Remember that all bits are paid for by the customers on both sides.
If, suddenly, your business model breaks because of a huge demand for high bandwidth services by your customers then either you need to charge your customers more or Netflix (or whoever) need to share the pie.
Anyone who builds an internet-related business model without taking into account the bandwidth use graph going straight from bottom left to upper right has no business being in this business. So charge your customers more if you have to. But charging third parties for the privilige of being able to send them the data your customers, who are already paying you, are trying to retrieve will turn the internet into the next cable or phone network where innovation happens within the confines of what the network owners feel comfortable with. In other words, the end of the internet as we know and love/hate it today. Don't slaughter the goose with the golden eggs.
On May 18, 2011, at 3:06 AM, Leigh Porter wrote:
-----Original Message----- From: Carl Rosevear [mailto:crosevear@skytap.com]
"Eating Up" sounds so overweight and unhealthy. Since a good number of us get paid for delivering bits, isn't this a good thing? Always glad to see bits and dollars flowing into the Internet, personally. However must express severe dissatisfaction with the topic of the thread a while ago referencing Comcast trying to charge providers for delivery over their network. Maybe I'm wrong, but I'm pretty happy with the current model... even if it means a $5/month residential rate hike (or something).
--C
Well it depends if Netflix pay for the bandwidth they use or if they get it all for free with non settlement peering. If, suddenly, your business model breaks because of a huge demand for high bandwidth services by your customers then either you need to charge your customers more or Netflix (or whoever) need to share the pie.
Netflix is hosted in ec2 and they use a lot of CDN. not sure that it's germain to the question of access to customers to measure which direction the money changes hands
-- Leigh Porter
______________________________________________________________________ This email has been scanned by the MessageLabs Email Security System. For more information please visit http://www.messagelabs.com/email ______________________________________________________________________
Since a good number of us get paid for delivering bits, isn't this a good thing?
at layer eight, having a single very large customer can be a source of unhappy surprises. randy
----- Original Message -----
From: "Randy Bush" <randy@psg.com>
Since a good number of us get paid for delivering bits, isn't this a good thing?
at layer eight, having a single very large customer can be a source of unhappy surprises.
I have first hand experience, having been laid off from my last IT director job because such a monopsony customer yanked 3/5 of its business from my then employer. Or ask *hundreds* of 35 year old companies that used to produce, nearly exclusively, lots of specialized, flight certified parts for the Space Shuttle Program. Cheers, -- jra -- Jay R. Ashworth Baylink jra@baylink.com Designer The Things I Think RFC 2100 Ashworth & Associates http://baylink.pitas.com 2000 Land Rover DII St Petersburg FL USA http://photo.imageinc.us +1 727 647 1274
Somebody should invent a a way to stream groups of shows simultaneously and just arrange for people to watch the desired stream at a particular time. Heck, maybe even do it wireless. problem solved, right? Cheers, Michael Holstein Cleveland State University
On Wed, May 18, 2011 at 12:46 PM, Michael Holstein < michael.holstein@csuohio.edu> wrote:
Somebody should invent a a way to stream groups of shows simultaneously and just arrange for people to watch the desired stream at a particular time. Heck, maybe even do it wireless.
problem solved, right?
There was a lengthy discussion about that on NANOG a week or so ago. I don't claim to understand all facets of multicast but it could be a sort of way to operate "tv station" type scheduled programming for streaming media. There's no way to pause, rewind or otherwise seek multicasted media though. It would be going backwards in terms of what consumers want these days. http://en.wikipedia.org/wiki/Multicast http://en.wikipedia.org/wiki/Mbone It seems to me that every provider these days is using a year 2K business model with 2011 bandwidth requirements and then complaining that consumers are transferring too much data. -- Landon Stewart <LStewart@SUPERB.NET> SuperbHosting.Net by Superb Internet Corp. Toll Free (US/Canada): 888-354-6128 x 4199 Direct: 206-438-5879 Web hosting and more "Ahead of the Rest": http://www.superbhosting.net
On Wed, May 18, 2011 at 3:56 PM, Landon Stewart <lstewart@superb.net> wrote:
There was a lengthy discussion about that on NANOG a week or so ago. I don't claim to understand all facets of multicast but it could be a sort of way to operate "tv station" type scheduled programming for streaming media. There's no way to pause, rewind or otherwise seek multicasted media though. It would be going backwards in terms of what consumers want these days.
why not permit your users to subscribe to shows/instances, stream them on-demand for viewing later... and leave truly live content (news/sports/etc) as is, with only the ability to pause/rewind? how is this different from broadcast tv today though?
On Wed, May 18, 2011 at 1:02 PM, Christopher Morrow <morrowc.lists@gmail.com> wrote:
On Wed, May 18, 2011 at 3:56 PM, Landon Stewart <lstewart@superb.net> wrote:
There was a lengthy discussion about that on NANOG a week or so ago. I don't claim to understand all facets of multicast but it could be a sort of way to operate "tv station" type scheduled programming for streaming media. There's no way to pause, rewind or otherwise seek multicasted media though. It would be going backwards in terms of what consumers want these days.
why not permit your users to subscribe to shows/instances, stream them on-demand for viewing later... and leave truly live content (news/sports/etc) as is, with only the ability to pause/rewind?
how is this different from broadcast tv today though?
It's not. These people need a pair of rabbit ears and a DVR. CB
On 18/05/11 1:13 PM, Cameron Byrne wrote:
It's not. These people need a pair of rabbit ears and a DVR.
Roughly 90% of the content I'm interested in watching is not available over the air. E.g. Comedy Central, CNN, Discovery, Showtime/HBO, etc. jc
On Wed, May 18, 2011 at 7:35 PM, JC Dill <jcdill.lists@gmail.com> wrote:
On 18/05/11 1:13 PM, Cameron Byrne wrote:
It's not. These people need a pair of rabbit ears and a DVR.
Roughly 90% of the content I'm interested in watching is not available over the air. E.g. Comedy Central, CNN, Discovery, Showtime/HBO, etc.
jc
Sure, but I'm guessing that something like that 80% of the content that 80% of people watch *is* available on some satellite/cable channel.
IP is perfect for the long tail, and yes, some of are mostly consumers of the tail :), but still, there is a win to be had on the front end of the beast...
On 18/05/11 4:42 PM, Dorn Hetzel wrote:
On Wed, May 18, 2011 at 7:35 PM, JC Dill <jcdill.lists@gmail.com <mailto:jcdill.lists@gmail.com>> wrote:
On 18/05/11 1:13 PM, Cameron Byrne wrote:
It's not. These people need a pair of rabbit ears and a DVR.
Roughly 90% of the content I'm interested in watching is not available over the air. E.g. Comedy Central, CNN, Discovery, Showtime/HBO, etc.
jc
Sure, but I'm guessing that something like that 80% of the content that 80% of people watch *is* available on some satellite/cable channel.
Yes, but most isn't available "over the air" with rabbit ears and a DVR. One of the big appeals of Netflix is the $8/month for all you can eat versus ~$40-60 for various cable and satellite packages. jc
Sure, but I'm guessing that something like that 80% of the content that 80% of people watch *is* available on some satellite/cable channel.
Yes, but most isn't available "over the air" with rabbit ears and a DVR. One of the big appeals of Netflix is the $8/month for all you can eat versus ~$40-60 for various cable and satellite packages.
jc
But it's not really $8/month, it's 8$ plus broadband.
On 18/05/11 5:10 PM, Dorn Hetzel wrote:
Sure, but I'm guessing that something like that 80% of the content that 80% of people watch *is* available on some satellite/cable channel.
Yes, but most isn't available "over the air" with rabbit ears and a DVR. One of the big appeals of Netflix is the $8/month for all you can eat versus ~$40-60 for various cable and satellite packages.
jc
But it's not really $8/month, it's 8$ plus broadband.
But I have broadband already. To get Satellite or Cable it's another $40-60/month, to get Netflix it's another $8/month. jc
why not permit your users to subscribe to shows/instances, stream them on-demand for viewing later... and leave truly live content (news/sports/etc) as is, with only the ability to pause/rewind?
how is this different from broadcast tv today though?
for some of us, the thing that is wonderful about netflix is the long tail. my tastes are a sigma or three out. randy
On Wed, May 18, 2011 at 4:15 PM, Randy Bush <randy@psg.com> wrote:
why not permit your users to subscribe to shows/instances, stream them on-demand for viewing later... and leave truly live content (news/sports/etc) as is, with only the ability to pause/rewind?
how is this different from broadcast tv today though?
for some of us, the thing that is wonderful about netflix is the long tail. my tastes are a sigma or three out.
usenet is over -> in all seriousness, if the content was available and you could request it be streamed to you 'sometime tomorrow' or 'sometime before Friday', you and the other people like you coudl get serviced on a singular 'stream'. I suspect that the vast majority of content is in the 1st sigma... and again, servicing everyone with a limited number of multicast'd streams seems like it would be nice. even falling back to unicast for some set of mathematically/cost-conscious examples seems like a win here.
for some of us, the thing that is wonderful about netflix is the long tail. my tastes are a sigma or three out. in all seriousness, if the content was available and you could request it be streamed to you 'sometime tomorrow' or 'sometime before Friday', you and the other people like you coudl get serviced on a singular 'stream'.
they do that now. by a station wagon full of holerith cards. well, how about a dvd in the post? randy
I think most the points made here are valid about why it isn't an easy problem to solve with multicast. Lets say for instance they had a multicast stream that sent the most popular content (which to Randy's point may not cover much) and 48 hours of that stream was cached locally on the CPE. What is the additional cost to expand each of these CPE's to handle this? Will it be HD or SD or both? Are people willing to Sacrafice their Xbox and PS3 disk space? Does the $60 Roku become the $400 Roku? Does securing all the content then become more difficult? What is the hard drive failure rate of these devices with them constantly writing to disk? What incentive do users have to to shell out the money for a device that will handle this caching? Multicasting this type of content seems to create more problems than it solves. On Wed, May 18, 2011 at 4:15 PM, Randy Bush <randy@psg.com> wrote:
why not permit your users to subscribe to shows/instances, stream them on-demand for viewing later... and leave truly live content (news/sports/etc) as is, with only the ability to pause/rewind?
how is this different from broadcast tv today though?
for some of us, the thing that is wonderful about netflix is the long tail. my tastes are a sigma or three out.
randy
From: Rick Astley [mailto:jnanog@gmail.com] I think most the points made here are valid about why it isn't an easy problem to solve with multicast. Lets say for instance they had a multicast stream that sent the most popular content (which to Randy's point may not cover much) and 48 hours of that stream was cached locally on the CPE. What is the additional cost to expand each of these CPE's to handle this? Will it be HD or SD or both? Are people willing to Sacrafice their Xbox and PS3 disk space? Does the $60 Roku become the $400 Roku? Does securing all the content then become more difficult? What is the hard drive failure rate of these devices with them constantly writing to disk?
What incentive do users have to to shell out the money for a device that will handle this caching? Multicasting this type of content seems to create more problems than it solves.
Lots of people already cache multicast streams to disk at home. I have a Humax digital TV cache (PVR ;-) that caches HD and SD content for me automatically. Doing the same over a network is not that much more of a jump really. My Humax box already has Ethernet to my home network, grabbing a multicast feed is no more than a software feature. So in a way people already pay to do just this. Indeed, in the UK, SKY offer a movies service which I believe you can cache locally if you have a SKY+ thing. So, SKY do it now and people pay for it. -- Leigh ______________________________________________________________________ This email has been scanned by the MessageLabs Email Security System. For more information please visit http://www.messagelabs.com/email ______________________________________________________________________
----- Original Message -----
From: "Christopher Morrow" <morrowc.lists@gmail.com>
why not permit your users to subscribe to shows/instances, stream them on-demand for viewing later... and leave truly live content (news/sports/etc) as is, with only the ability to pause/rewind?
how is this different from broadcast tv today though?
It's on the Internet. So it's cooler. Cheers, -- jra -- Jay R. Ashworth Baylink jra@baylink.com Designer The Things I Think RFC 2100 Ashworth & Associates http://baylink.pitas.com 2000 Land Rover DII St Petersburg FL USA http://photo.imageinc.us +1 727 647 1274
I think this shows the need for an Internet-wide multicast implementation. Although I can recall working on a product that delivered satellite multicast streams (with each multicast group corresponding to individual TV stations) to telco CO's. This enabled the telco to implement multicast at the edge of their networks, where user broadband clients would issue multicast joins only as far as the CO. If I recall this was implemented with the old Cincinnati Bell telco. I admit there are a lot of CO's and cable head-ends though for this solution to scale. -----Original Message----- From: Michael Holstein [mailto:michael.holstein@csuohio.edu] Sent: Wednesday, May 18, 2011 12:46 PM To: Roy Cc: nanog Subject: Re: Netflix Is Eating Up More Of North America's Bandwidth Than Any Other Company
Somebody should invent a a way to stream groups of shows simultaneously and just arrange for people to watch the desired stream at a particular time. Heck, maybe even do it wireless. problem solved, right? Cheers, Michael Holstein Cleveland State University This communication, together with any attachments or embedded links, is for the sole use of the intended recipient(s) and may contain information that is confidential or legally protected. If you are not the intended recipient, you are hereby notified that any review, disclosure, copying, dissemination, distribution or use of this communication is strictly prohibited. If you have received this communication in error, please notify the sender immediately by return e-mail message and delete the original and all copies of the communication, along with any attachments or embedded links, from your system.
On 2011-05-18, at 16:01, Holmes,David A wrote:
I think this shows the need for an Internet-wide multicast implementation.
Or perhaps even some kind of new technology that is independent of the Internet! Imagine such futuristic ideas as solar-powered spacecraft in orbit around the planet bouncing content back across massive areas so that everybody can pick them up at once. Crazy stuff. Joe
Joe Abley wrote:
Or perhaps even some kind of new technology that is independent of the Internet! Imagine such futuristic ideas as solar-powered spacecraft in orbit around the planet bouncing content back across massive areas so that everybody can pick them up at once.
Crazy stuff.
You mean like a sputnik? Crazy indeed... -- http://goldmark.org/jeff/stupid-disclaimers/ http://linuxmafia.com/~rick/faq/plural-of-virus.html
On Wed, May 18, 2011 at 4:27 PM, Jeroen van Aart <jeroen@mompl.net> wrote:
Joe Abley wrote:
Or perhaps even some kind of new technology that is independent of the Internet! Imagine such futuristic ideas as solar-powered spacecraft in orbit around the planet bouncing content back across massive areas so that everybody can pick them up at once.
Crazy stuff.
You mean like a sputnik?
sputnik was VERY low bandwidth though... if you wanted to stream a current movie, you'd likely have to have started when sputnik actually launched to be sure you'd be able to watch it next year.
On 05/18/2011 04:01 PM, Holmes,David A wrote:
I think this shows the need for an Internet-wide multicast implementation. Although I can recall working on a product that delivered satellite multicast streams (with each multicast group corresponding to individual TV stations) to telco CO's. This enabled the telco to implement multicast at the edge of their networks, where user broadband clients would issue multicast joins only as far as the CO. If I recall this was implemented with the old Cincinnati Bell telco. I admit there are a lot of CO's and cable head-ends though for this solution to scale.
I don't see how multicast necasarily solves the netflix on-demand video problem. you have millions of users streaming different content at different times. multicast is great for the world cup but how does it solve the video on demand problem?
I don't see how multicast necasarily solves the netflix on-demand video problem. you have millions of users streaming different content at different times. multicast is great for the world cup but how does it solve the video on demand problem?
I suppose in theory if you have tivo-like devices at the endpoints then
they can capture popular programs at the time of multicast for later viewing. Whether this is better than capturing the same programs over a broadcast medium for later playback, I don't know...
On 2011-05-18, at 16:09, Dorn Hetzel wrote:
they can capture popular programs at the time of multicast for later viewing. Whether this is better than capturing the same programs over a broadcast medium for later playback, I don't know...
... or a peer to peer medium, which is (as I understand it) how people who really want this to happen today manage to do it. The problem is not the distribution so much as the need to shoe-horn this network efficiency into the content business model. I heard similar stories about the early days of distributing digital copies of movies to theatres for presentation -- the technology was trivial, even with fairly low-power commodity CPUs, until you insist that the content be encrypted so that nobody can walk off with it without paying. Joe
On May 18, 2011, at 1:01 PM, Holmes,David A wrote:
I think this shows the need for an Internet-wide multicast implementation.
there's a pretty longtailed distribution on what people might chose to stream. static content is ameniable to distribution via cdn (which is frankly a degenerate form of multicast), but lets face it, how many people watched "Charles Mingus: Triumph of the Underdog" in east palo alto last night at 10pm.
Although I can recall working on a product that delivered satellite multicast streams (with each multicast group corresponding to individual TV stations) to telco CO's. This enabled the telco to implement multicast at the edge of their networks, where user broadband clients would issue multicast joins only as far as the CO. If I recall this was implemented with the old Cincinnati Bell telco. I admit there are a lot of CO's and cable head-ends though for this solution to scale.
-----Original Message----- From: Michael Holstein [mailto:michael.holstein@csuohio.edu] Sent: Wednesday, May 18, 2011 12:46 PM To: Roy Cc: nanog Subject: Re: Netflix Is Eating Up More Of North America's Bandwidth Than Any Other Company
Somebody should invent a a way to stream groups of shows simultaneously and just arrange for people to watch the desired stream at a particular time. Heck, maybe even do it wireless.
problem solved, right?
Cheers,
Michael Holstein Cleveland State University
This communication, together with any attachments or embedded links, is for the sole use of the intended recipient(s) and may contain information that is confidential or legally protected. If you are not the intended recipient, you are hereby notified that any review, disclosure, copying, dissemination, distribution or use of this communication is strictly prohibited. If you have received this communication in error, please notify the sender immediately by return e-mail message and delete the original and all copies of the communication, along with any attachments or embedded links, from your system.
On Wed, May 18, 2011 at 4:18 PM, Joel Jaeggli <joelja@bogus.com> wrote:
On May 18, 2011, at 1:01 PM, Holmes,David A wrote:
I think this shows the need for an Internet-wide multicast implementation.
there's a pretty longtailed distribution on what people might chose to stream. static content is ameniable to distribution via cdn (which is frankly a degenerate form of multicast), but lets face it, how many people watched "Charles Mingus: Triumph of the Underdog" in east palo alto last night at 10pm.
slightly wrong question: "How many people last 'period of time' chose, early enough, to want to watch CMTotU lastnight at 10." if the number is greater than X, multicast it with time to deliver before 10pm pdt start time. If it's less, unicast...
If we're really talking efficiency, the "popular" stuff should probably stream out over the bird of your choice (directv, etc) because it's hard to beat millions of dishes and dvr's and no cable plant. Then what won't fit on the bird goes unicast IP from the nearest CDN. Kind of like the "on demand over broadband" on my satellite box. Their selection sucks, but the model is valid. On Wed, May 18, 2011 at 4:18 PM, Joel Jaeggli <joelja@bogus.com> wrote:
On May 18, 2011, at 1:01 PM, Holmes,David A wrote:
I think this shows the need for an Internet-wide multicast implementation.
there's a pretty longtailed distribution on what people might chose to stream. static content is ameniable to distribution via cdn (which is frankly a degenerate form of multicast), but lets face it, how many people watched "Charles Mingus: Triumph of the Underdog" in east palo alto last night at 10pm.
Although I can recall working on a product that delivered satellite multicast streams (with each multicast group corresponding to individual TV stations) to telco CO's. This enabled the telco to implement multicast at the edge of their networks, where user broadband clients would issue multicast joins only as far as the CO. If I recall this was implemented with the old Cincinnati Bell telco. I admit there are a lot of CO's and cable head-ends though for this solution to scale.
-----Original Message----- From: Michael Holstein [mailto:michael.holstein@csuohio.edu] Sent: Wednesday, May 18, 2011 12:46 PM To: Roy Cc: nanog Subject: Re: Netflix Is Eating Up More Of North America's Bandwidth Than Any Other Company
Somebody should invent a a way to stream groups of shows simultaneously and just arrange for people to watch the desired stream at a particular time. Heck, maybe even do it wireless.
problem solved, right?
Cheers,
Michael Holstein Cleveland State University
This communication, together with any attachments or embedded links, is for the sole use of the intended recipient(s) and may contain information that is confidential or legally protected. If you are not the intended recipient, you are hereby notified that any review, disclosure, copying, dissemination, distribution or use of this communication is strictly prohibited. If you have received this communication in error, please notify the sender immediately by return e-mail message and delete the original and all copies of the communication, along with any attachments or embedded links, from your system.
On 5/18/11 2:33 PM, Dorn Hetzel wrote:
If we're really talking efficiency, the "popular" stuff should probably stream out over the bird of your choice (directv, etc) because it's hard to beat millions of dishes and dvr's and no cable plant.
Then what won't fit on the bird goes unicast IP from the nearest CDN. Kind of like the "on demand over broadband" on my satellite box. Their selection sucks, but the model is valid.
If someone hadn't mentioned already, there used to be a usenet provider that delivered a full feed via Satellite. Anything is feasible, just have to find people who actually want/need it and a provider that isn't blind to long term benefits. -- Brielle Bruns The Summit Open Source Development Group http://www.sosdg.org / http://www.ahbl.org
On Wed, May 18, 2011 at 4:53 PM, Brielle Bruns <bruns@2mbit.com> wrote:
On 5/18/11 2:33 PM, Dorn Hetzel wrote:
If we're really talking efficiency, the "popular" stuff should probably stream out over the bird of your choice (directv, etc) because it's hard to beat millions of dishes and dvr's and no cable plant.
Then what won't fit on the bird goes unicast IP from the nearest CDN. Kind of like the "on demand over broadband" on my satellite box. Their selection sucks, but the model is valid.
If someone hadn't mentioned already, there used to be a usenet provider that delivered a full feed via Satellite. Anything is feasible, just have to
doug went out of that business, it wasn't (apparently) actually viable.
On Wed, 18 May 2011, Brielle Bruns wrote:
If someone hadn't mentioned already, there used to be a usenet provider that delivered a full feed via Satellite. Anything is feasible, just have to find people who actually want/need it and a provider that isn't blind to long term benefits.
Skycache/Cidera...until it didn't fit anymore in the bandwidth they had. IIRC, it was only around 28mbps. Also, IIRC, that "business" was a sort of after thought after their original plan (squid cache pre-population) didn't pan out. Anyone want to buy some Skycache chopsticks? I think I still have a few unopened sets from whichever late 90s ISPCon I went to in San Jose, CA...Skycache rented out some museum for a sushi party. ---------------------------------------------------------------------- Jon Lewis, MCP :) | I route Senior Network Engineer | therefore you are Atlantic Net | _________ http://www.lewis.org/~jlewis/pgp for PGP public key_________
If someone hadn't mentioned already, there used to be a usenet provider
There was also "Planet Connect" years ago that delivered full Usenet (128K worth) along with all my Fidonet BBS updates too .. I think I just dated myself ;) We still have an old Cidera system on a rooftop that nobody has taken down yet ... Paul -----Original Message----- From: Jon Lewis [mailto:jlewis@lewis.org] Sent: May-18-11 6:01 PM To: Brielle Bruns Cc: nanog@nanog.org Subject: Re: Netflix Is Eating Up More Of North America's Bandwidth Than Any Other Company On Wed, 18 May 2011, Brielle Bruns wrote: that
delivered a full feed via Satellite. Anything is feasible, just have to find people who actually want/need it and a provider that isn't blind to long term benefits.
Skycache/Cidera...until it didn't fit anymore in the bandwidth they had. IIRC, it was only around 28mbps. Also, IIRC, that "business" was a sort of after thought after their original plan (squid cache pre-population) didn't pan out. Anyone want to buy some Skycache chopsticks? I think I still have a few unopened sets from whichever late 90s ISPCon I went to in San Jose, CA...Skycache rented out some museum for a sushi party. ---------------------------------------------------------------------- Jon Lewis, MCP :) | I route Senior Network Engineer | therefore you are Atlantic Net | _________ http://www.lewis.org/~jlewis/pgp for PGP public key_________
From nanog-bounces+bonomi=mail.r-bonomi.com@nanog.org Wed May 18 16:12:17 2011 Date: Wed, 18 May 2011 14:53:10 -0600 From: Brielle Bruns <bruns@2mbit.com> To: nanog@nanog.org Subject: Re: Netflix Is Eating Up More Of North America's Bandwidth Than Any Other Company
On 5/18/11 2:33 PM, Dorn Hetzel wrote:
If we're really talking efficiency, the "popular" stuff should probably stream out over the bird of your choice (directv, etc) because it's hard to beat millions of dishes and dvr's and no cable plant.
Then what won't fit on the bird goes unicast IP from the nearest CDN. Kind of like the "on demand over broadband" on my satellite box. Their selection sucks, but the model is valid.
If someone hadn't mentioned already, there used to be a usenet provider that delivered a full feed via Satellite.
There were, at different times, two companies that did that. Both went under because expenses exceeded income. The one that was _much_ more interesting was the one that Lauren Weinstein had a hand in. It piggy-backed a Usenet feed in the vertical blanking interval of several big "independant" TV stations -- ones that were carried by practically every cable company in the country. Distribution to the cable companies was via satellite, but the USENET feed, being _part_ of the video signal, consumed _zero_ additional bandwidth, and rode the satellite links for free. To get the feed, all you needed was a TV tuner with 'video out', and the purpose-huilt decoder box that extracted the vertical interval data. This service died as the independants moved to encrypted transmission, because the encryption did _not_ perserve anything in the 'blanking' timeslot. only the 'viewable' field-image was trasmitted, _as_ a full-field image. Sync, blanking, etc. was _locally_ generated on the receiving end. An "elegant" idea, done in by changing technology. *sigh*
Was PBS one of the companies you are referring to? A colleague of mine worked as a developer on a project at PBS in the 90s that used the blanking interval for Internet transmissio - very cool stuff. On 5/19/11, Robert Bonomi <bonomi@mail.r-bonomi.com> wrote:
From nanog-bounces+bonomi=mail.r-bonomi.com@nanog.org Wed May 18 16:12:17 2011 Date: Wed, 18 May 2011 14:53:10 -0600 From: Brielle Bruns <bruns@2mbit.com> To: nanog@nanog.org Subject: Re: Netflix Is Eating Up More Of North America's Bandwidth Than Any Other Company
On 5/18/11 2:33 PM, Dorn Hetzel wrote:
If we're really talking efficiency, the "popular" stuff should probably stream out over the bird of your choice (directv, etc) because it's hard to beat millions of dishes and dvr's and no cable plant.
Then what won't fit on the bird goes unicast IP from the nearest CDN. Kind of like the "on demand over broadband" on my satellite box. Their selection sucks, but the model is valid.
If someone hadn't mentioned already, there used to be a usenet provider that delivered a full feed via Satellite.
There were, at different times, two companies that did that. Both went under because expenses exceeded income.
The one that was _much_ more interesting was the one that Lauren Weinstein had a hand in. It piggy-backed a Usenet feed in the vertical blanking interval of several big "independant" TV stations -- ones that were carried by practically every cable company in the country. Distribution to the cable companies was via satellite, but the USENET feed, being _part_ of the video signal, consumed _zero_ additional bandwidth, and rode the satellite links for free.
To get the feed, all you needed was a TV tuner with 'video out', and the purpose-huilt decoder box that extracted the vertical interval data.
This service died as the independants moved to encrypted transmission, because the encryption did _not_ perserve anything in the 'blanking' timeslot. only the 'viewable' field-image was trasmitted, _as_ a full-field image. Sync, blanking, etc. was _locally_ generated on the receiving end.
An "elegant" idea, done in by changing technology. *sigh*
On Tue, May 24, 2011 at 08:12:31PM -0400, Max wrote:
Was PBS one of the companies you are referring to? A colleague of mine worked as a developer on a project at PBS in the 90s that used the blanking interval for Internet transmissio - very cool stuff.
<snip>
The one that was _much_ more interesting was the one that Lauren Weinstein had a hand in. It piggy-backed a Usenet feed in the vertical blanking interval of several big "independant" TV stations -- ones that were carried by practically every cable company in the country. Distribution to the cable companies was via satellite, but the USENET feed, being _part_ of the video signal, consumed _zero_ additional bandwidth, and rode the satellite links for free.
To get the feed, all you needed was a TV tuner with 'video out', and the purpose-huilt decoder box that extracted the vertical interval data.
This service died as the independants moved to encrypted transmission, because the encryption did _not_ perserve anything in the 'blanking' timeslot. only the 'viewable' field-image was trasmitted, _as_ a full-field image. Sync, blanking, etc. was _locally_ generated on the receiving end.
An "elegant" idea, done in by changing technology. *sigh*
As USENIX director I sponsored and sheparded this project, called "Stargate". We at least got bits into the blanking interval at WTBS in Altanta. -- -=[L]=- Hand typed on my Remington portable Real data are normal in the middle and Cauchy in the tails.
On Tue, May 24, 2011 at 10:48 PM, Lou Katz <lou@metron.com> wrote:
An "elegant" idea, done in by changing technology. *sigh*
As USENIX director I sponsored and sheparded this project, called "Stargate". We at least got bits into the blanking interval at WTBS in Altanta.
So... would this have been feasible today? given the bandwidth required to send a full feed these days, i suspect likely not, eh? (even if you were able to do it on all 500+ channels in parallel)
----- Original Message -----
From: "Christopher Morrow" <morrowc.lists@gmail.com>
On Tue, May 24, 2011 at 10:48 PM, Lou Katz <lou@metron.com> wrote:
An "elegant" idea, done in by changing technology. *sigh*
As USENIX director I sponsored and sheparded this project, called "Stargate". We at least got bits into the blanking interval at WTBS in Altanta.
So... would this have been feasible today? given the bandwidth required to send a full feed these days, i suspect likely not, eh? (even if you were able to do it on all 500+ channels in parallel)
I can't tell you whether it would be feasible from a *quantity* standpoint unless you specify what your group list is -- big 7 text? Probably. Problem is, it depended (as he noted) on a peculiarity of the network TV environment at the time: it wasn't part of the signal, but of the *transport* which -- at the time -- was carried around along with the signal, so you could piggyback stuff there, and get it right to people's TVs. MPEG2 and 4 don't carry the vertical interval, so any ride you can get isn't free -- rather similar to our Multicast discussion last week. Back in the really bad old days, I'm told that the most stable frequency source the average civilian could get was the 3.58MHz oscillator in a color TV set -- but *only* when you were watching *network* programs, at which time that oscillator was effectively phase-locked to a $50k+ black burst generator at network master control. Frame synchronizers shot that plan out of the water. Never been sure if that's apocryphal or not. Cheers, -- jra -- Jay R. Ashworth Baylink jra@baylink.com Designer The Things I Think RFC 2100 Ashworth & Associates http://baylink.pitas.com 2000 Land Rover DII St Petersburg FL USA http://photo.imageinc.us +1 727 647 1274
From nanog-bounces+bonomi=mail.r-bonomi.com@nanog.org Tue May 24 22:19:18 2011 Date: Tue, 24 May 2011 23:14:56 -0400 (EDT) From: Jay Ashworth <jra@baylink.com> To: NANOG <nanog@nanog.org> Subject: Re: Netflix Is Eating Up More Of North America's Bandwidth Than Any Other Company
----- Original Message -----
From: "Christopher Morrow" <morrowc.lists@gmail.com>
On Tue, May 24, 2011 at 10:48 PM, Lou Katz <lou@metron.com> wrote:
An "elegant" idea, done in by changing technology. *sigh*
As USENIX director I sponsored and sheparded this project, called "Stargate". We at least got bits into the blanking interval at WTBS in Altanta.
So... would this have been feasible today? given the bandwidth required to send a full feed these days, i suspect likely not, eh? (even if you were able to do it on all 500+ channels in parallel)
I can't tell you whether it would be feasible from a *quantity* standpoint unless you specify what your group list is -- big 7 text? Probably.
Problem is, it depended (as he noted) on a peculiarity of the network TV environment at the time: it wasn't part of the signal, but of the *transport* which -- at the time -- was carried around along with the signal, so you could piggyback stuff there, and get it right to people's TVs. MPEG2 and 4 don't carry the vertical interval, so any ride you can get isn't free -- rather similar to our Multicast discussion last week.
Back in the really bad old days, I'm told that the most stable frequency source the average civilian could get was the 3.58MHz oscillator in a color TV set -- but *only* when you were watching *network* programs, at which time that oscillator was effectively phase-locked to a $50k+ black burst generator at network master control.
Frame synchronizers shot that plan out of the water.
Never been sure if that's apocryphal or not.
Cheers, -- jra -- Jay R. Ashworth Baylink jra@baylink.com Designer The Things I Think RFC 2100 Ashworth & Associates http://baylink.pitas.com 2000 Land Rover DII St Petersburg FL USA http://photo.imageinc.us +1 727 647 1274
From nanog-bounces+bonomi=mail.r-bonomi.com@nanog.org Tue May 24 22:12:57 2011 Date: Tue, 24 May 2011 23:12:41 -0400 Subject: Re: Netflix Is Eating Up More Of North America's Bandwidth Than Any Other Company From: Christopher Morrow <morrowc.lists@gmail.com> To: nanog@nanog.org
On Tue, May 24, 2011 at 10:48 PM, Lou Katz <lou@metron.com> wrote:
An "elegant" idea, done in by changing technology. *sigh*
As USENIX director I sponsored and sheparded this project, called "Stargate". We at least got bits into the blanking interval at WTBS in Altanta.
So... would this have been feasible today? given the bandwidth required to send a full feed these days, i suspect likely not, eh? (even if you were able to do it on all 500+ channels in parallel)
On the financial side, it is trivial. On the engineering side, _impossible_. Modern satellite feeds of NTSC (analog) TV signals use compressed digital representations of only the image portion of each 'field' of the video stream. Sync and blanking signals are _not_ transmitted; rather they are re-generated locally when the satellite stream is converted back to an NTSC analog signal. Thus, even if you were to inject data in the vertical interval, it would be stripped before satellite uplink, and not recoverable at the receiving side. It's been a *LONG* time since I dealt with the data bandwith available in the vertical interval, but, as I recall, the "raw' capacity is on the same order as a DS-0. *but* you have to deduct overhead for framing, FEC, etc, as well as multiple redundant transmissions of each data 'packet'. To pick a conservative number, say you get an effective throughput of 2k bytes/sec, That is roughly 150mbyte/day. That _might_ suffice for a text-only, 'Big-8' only, feed.. As I understand it, a current USENET 'full feed', including binaries, take two dedicated 100mbit FDX fast ethernet links, and they are saturated _most_ of the day. At that rate, A full day of TV vertical interval transmission wuould handle under _ten_seconds_ worth of the inbound traffic. You would around =ten=thousand= analog TV channels to handle a contemporary 'full feed'.
----- Original Message -----
From: "Robert Bonomi" <bonomi@mail.r-bonomi.com>
As I understand it, a current USENET 'full feed', including binaries, take two dedicated 100mbit FDX fast ethernet links, and they are saturated _most_ of the day. At that rate, A full day of TV vertical interval transmission wuould handle under _ten_seconds_ worth of the inbound traffic. You would around =ten=thousand= analog TV channels to handle a contemporary 'full feed'.
And I remember the month, August 1982 I think it was, when I stopped being able to read *all* of Usenet. At least, everything my junior college took. :-) Cheers, - jra -- Jay R. Ashworth Baylink jra@baylink.com Designer The Things I Think RFC 2100 Ashworth & Associates http://baylink.pitas.com 2000 Land Rover DII St Petersburg FL USA http://photo.imageinc.us +1 727 647 1274
On 05/24/2011 11:12 PM, Christopher Morrow wrote:
On Tue, May 24, 2011 at 10:48 PM, Lou Katz<lou@metron.com> wrote:
An "elegant" idea, done in by changing technology. *sigh*
As USENIX director I sponsored and sheparded this project, called "Stargate". We at least got bits into the blanking interval at WTBS in Altanta.
So... would this have been feasible today? given the bandwidth required to send a full feed these days, i suspect likely not, eh? (even if you were able to do it on all 500+ channels in parallel)
All you need these days is an MPEG4 decoder and some ATSC tuners, and you can regenerate most of the alt. namespace locally. For the rest you may need a subscription to netflix a DVD drive, and some ripping software. I wonder what usenet would take for bandwidth if you ditched all the pirated content. -Paul
It was TBS, in the 1980s: http://web.archive.org/web/19981203103811/www.stargate.com/history.html It used TBS because that was one of the first "superstations", distributed to cable systems nationwide via satellite. On May 24, 2011, at 8:12 31PM, Max wrote:
Was PBS one of the companies you are referring to? A colleague of mine worked as a developer on a project at PBS in the 90s that used the blanking interval for Internet transmissio - very cool stuff.
On 5/19/11, Robert Bonomi <bonomi@mail.r-bonomi.com> wrote:
From nanog-bounces+bonomi=mail.r-bonomi.com@nanog.org Wed May 18 16:12:17 2011 Date: Wed, 18 May 2011 14:53:10 -0600 From: Brielle Bruns <bruns@2mbit.com> To: nanog@nanog.org Subject: Re: Netflix Is Eating Up More Of North America's Bandwidth Than Any Other Company
On 5/18/11 2:33 PM, Dorn Hetzel wrote:
If we're really talking efficiency, the "popular" stuff should probably stream out over the bird of your choice (directv, etc) because it's hard to beat millions of dishes and dvr's and no cable plant.
Then what won't fit on the bird goes unicast IP from the nearest CDN. Kind of like the "on demand over broadband" on my satellite box. Their selection sucks, but the model is valid.
If someone hadn't mentioned already, there used to be a usenet provider that delivered a full feed via Satellite.
There were, at different times, two companies that did that. Both went under because expenses exceeded income.
The one that was _much_ more interesting was the one that Lauren Weinstein had a hand in. It piggy-backed a Usenet feed in the vertical blanking interval of several big "independant" TV stations -- ones that were carried by practically every cable company in the country. Distribution to the cable companies was via satellite, but the USENET feed, being _part_ of the video signal, consumed _zero_ additional bandwidth, and rode the satellite links for free.
To get the feed, all you needed was a TV tuner with 'video out', and the purpose-huilt decoder box that extracted the vertical interval data.
This service died as the independants moved to encrypted transmission, because the encryption did _not_ perserve anything in the 'blanking' timeslot. only the 'viewable' field-image was trasmitted, _as_ a full-field image. Sync, blanking, etc. was _locally_ generated on the receiving end.
An "elegant" idea, done in by changing technology. *sigh*
--Steve Bellovin, https://www.cs.columbia.edu/~smb
On Tue, May 24, 2011 at 10:48 PM, Steven Bellovin <smb@cs.columbia.edu> wrote:
It was TBS, in the 1980s: http://web.archive.org/web/19981203103811/www.stargate.com/history.html
It used TBS because that was one of the first "superstations", distributed to cable systems nationwide via satellite.
oops - meant TBS :), that was it. - Max
See also: UK efforts http://en.wikipedia.org/wiki/Prestel <http://en.wikipedia.org/wiki/Prestel>j On Wed, May 25, 2011 at 12:04 AM, Max <perldork@webwizarddesign.com> wrote:
On Tue, May 24, 2011 at 10:48 PM, Steven Bellovin <smb@cs.columbia.edu> wrote:
It was TBS, in the 1980s: http://web.archive.org/web/19981203103811/www.stargate.com/history.html
It used TBS because that was one of the first "superstations", distributed to cable systems nationwide via satellite.
oops - meant TBS :), that was it.
- Max
-- --------------------------------------------------------------- Joly MacFie 218 565 9365 Skype:punkcast WWWhatsup NYC - http://wwwhatsup.com http://pinstand.com - http://punkcast.com VP (Admin) - ISOC-NY - http://isoc-ny.org -------------------------------------------------------------- -
----- Original Message -----
From: "Joel Jaeggli" <joelja@bogus.com>
On May 18, 2011, at 1:01 PM, Holmes,David A wrote:
I think this shows the need for an Internet-wide multicast implementation.
there's a pretty longtailed distribution on what people might chose to stream. static content is ameniable to distribution via cdn (which is frankly a degenerate form of multicast), but lets face it, how many people watched "Charles Mingus: Triumph of the Underdog" in east palo alto last night at 10pm.
Of course. But that's a strawman. What percentage of available titles, by *count*, accounts for even 50% of the streamed data, in bytes? 2%? 1? Cheers, -- jra -- Jay R. Ashworth Baylink jra@baylink.com Designer The Things I Think RFC 2100 Ashworth & Associates http://baylink.pitas.com 2000 Land Rover DII St Petersburg FL USA http://photo.imageinc.us +1 727 647 1274
On 19/05/2011, at 6:01 AM, "Holmes,David A" <dholmes@mwdh2o.com> wrote:
I think this shows the need for an Internet-wide multicast implementation. Although I can recall working on a product that delivered satellite multicast streams (with each multicast group corresponding to individual TV stations) to telco CO's. This enabled the telco to implement multicast at the edge of their networks, where user broadband clients would issue multicast joins only as far as the CO. If I recall this was implemented with the old Cincinnati Bell telco. I admit there are a lot of CO's and cable head-ends though for this solution to scale.
-----Original Message----- From: Michael Holstein [mailto:michael.holstein@csuohio.edu] Sent: Wednesday, May 18, 2011 12:46 PM To: Roy Cc: nanog Subject: Re: Netflix Is Eating Up More Of North America's Bandwidth Than Any Other Company
Somebody should invent a a way to stream groups of shows simultaneously and just arrange for people to watch the desired stream at a particular time. Heck, maybe even do it wireless.
problem solved, right?
Cheers,
Michael Holstein Cleveland State University
No matter where you go, there you are. [--anon?] or Those who don't understand history are doomed to repeat it. - [heavily paraphrased -- Santayana] jy
----- Original Message -----
From: "Jeffrey S. Young" <young@jsyoung.net>
Somebody should invent a a way to stream groups of shows simultaneously and just arrange for people to watch the desired stream at a particular time. Heck, maybe even do it wireless.
problem solved, right?
Those who don't understand history are doomed to repeat it. - [heavily paraphrased -- Santayana]
"Those who do not understand broadcasting are doomed to reinvent it. Poorly." --after Henry Spencer. Cheers, -- jra -- Jay R. Ashworth Baylink jra@baylink.com Designer The Things I Think RFC 2100 Ashworth & Associates http://baylink.pitas.com 2000 Land Rover DII St Petersburg FL USA http://photo.imageinc.us +1 727 647 1274
On Thu, May 19, 2011 at 07:45:44AM +1000, Jeffrey S. Young wrote:
No matter where you go, there you are. [--anon?]
Oliver's Law of Location Kinda usurped by Buckaroo Banzai in the movie by the same name. It always annoys me when attributed to that character. Go back to your regular programming -- multicast, unicast, or broadcast it'll proabably be more interesting this this dead end...
You mean IP TV content products from folks such as SES Americom' IP-PRIME or IPTV Americas or EchoStart IP TV or Intelsat? -----Original Message----- From: Holmes,David A [mailto:dholmes@mwdh2o.com] Sent: Wednesday, May 18, 2011 3:01 PM To: Michael Holstein; Roy Cc: nanog Subject: RE: Netflix Is Eating Up More Of North America's Bandwidth Than Any Other Company I think this shows the need for an Internet-wide multicast implementation. Although I can recall working on a product that delivered satellite multicast streams (with each multicast group corresponding to individual TV stations) to telco CO's. This enabled the telco to implement multicast at the edge of their networks, where user broadband clients would issue multicast joins only as far as the CO. If I recall this was implemented with the old Cincinnati Bell telco. I admit there are a lot of CO's and cable head-ends though for this solution to scale.
On Tue, May 17, 2011 at 5:02 PM, Roy <r.engehausen@gmail.com> wrote:
You know... I say the way the headline characterizes the subject is misleading. It would be more accurate to say something like.... "North American Internet users utilize more of their available network capacity to access Netflix, than any other company's service." Bandwidth is not "eaten up"; it is utilized. Bandwidth is not a scarce, indivisable resource. Since the article is about "peak downstream bandwidth in North America". The statistics are about what broadband customers are doing; according to Sandvine. The first thought that enters my mind, when I see this headline, and those charts, is that the provider of the statistics / chooser of the headline has a product they want to sell. :) -- -JH
participants (35)
-
Alex Brooks -
bmanning＠vacation.karoshi.com -
Brielle Bruns -
Cameron Byrne -
Carl Rosevear -
Christopher Morrow -
Dorn Hetzel -
Eliot Lear -
Frank Bulk -
Holmes,David A -
Iljitsch van Beijnum -
Jay Ashworth -
JC Dill -
Jeffrey S. Young -
Jeroen van Aart -
Jimmy Hess -
Joe Abley -
Joel Jaeggli -
John Osmon -
Joly MacFie -
Jon Lewis -
Landon Stewart -
Leigh Porter -
Lou Katz -
Matt Ryanczak -
Max -
Michael Holstein -
Paul Stewart -
Paul Timmins -
Phil Regnauld -
Randy Bush -
Rick Astley -
Robert Bonomi -
Roy -
Steven Bellovin