how-to block ads
|
Share Topic
 |
 |
|
|
|
 IPPlanManHoly Cable Modem Batman
join:2000-09-20
Washington, DC
kudos:1
2 edits | Cable internet may be the dying tehchology You said:
"Yes. Dying technology."
Based on Espaeth's most excellent post, it seems to me that Cable Internet might actually be the dying technology that you say DSL is. | |
| 
3 edits | Re: Cable internet may be the dying tehchology I agree, but since cable is already FTTN, it is easy and cheap to upgrade the: nodes, head ends and cables to go FTTH.
Check this out:
Method for upgrading and converting a coaxial cable with a fiber optic cable
Here is what they could use
»www.youtube.com/watch?v=mUBRjiVhJTs&fmt=18
Abstract:
A communications system including an initially installed coaxial cable may be readily upgraded by installing a fiber optic cable therein. The coaxial cable has an empty longitudinal channel extending continuously along substantially the entire length thereof. A propellable fiber optic cable is later installed in the channel by a fluid flow which creates a sufficient drag to advance the fiber optic cable into the channel. The fiber optic cable is installed to meet increasing communications demands or when otherwise desired. The coaxial cable may include an empty electrically conductive tube serving as the center conductor for later receipt therein of the fiber optic cable. In other embodiments, the coaxial cable may have a channel provided in a dielectric material between the center and outer conductors, or a channel formed in the outer protective jacket.
They pull the copper and metal out of the plastic sheath, and then shoot fiber through, pretty neat!
Also:
Method for upgrading a hybrid fiber coax network to an all fiber network
Abstract:
Disclosed is a method for making the transition from a hybrid network into an all fiber network. In the preferred embodiment, the first step is to replace an existing fiber optic/coaxial cable conversion device located at a remote node with an optic distribution device. Secondly, a house optical network unit is installed in a living unit. Finally, a downstream optic fiber is installed between the optic distribution device and the house optical network unit. | |
| | IPPlanManHoly Cable Modem Batman
join:2000-09-20
Washington, DC
kudos:1
1 edit | Re: Cable internet may be the dying tehchology Interesting... It may come to that.
Makes me wonder if Comcast has any similar ground-up plans. | |
| | |
1 edit | Re: Cable internet may be the dying tehchology Thats the great thing, its not ground up! they already have 80% of the infrastructure! | |
| | |  espaethDigital PlumberPremium,MVM
join:2001-04-21
Minneapolis, MN
kudos:2
 ·Vitelity VOIP
1 edit | said by IPPlanMan:Interesting... It may come to that. Makes me wonder if Comcast has any similar ground-up plans. Doubtful.
When you look at it, FiOS is really a HFC system with the only difference being the node is at each house. The handoff within each residence is a coax cable -- the TV is distributed within the home using standard QAM over coax delivery so you can use a TiVo or a VZ-provided cable box that can do on-demand. The handoff to the router can be either Cat5 from the ONT or it can also use the MoCA transport over coax.
For the cable system they can keep adding capacity without residential dispatch by continuing to reclaim analog channel space and performing node splits to reduce the number of homes per RF segment. The capacity of the DOCSIS plant can actually match or exceed that of GPON networks if they did the same 16 homes per node scaling, especially if they were aggressive in converting over to switched digital video in each of the segments.
The HFC cable plant is a long way from dead. | |
| | | | IPPlanManHoly Cable Modem Batman
join:2000-09-20
Washington, DC
kudos:1
3 edits | Again, opinion cap is meant to prevent competitive VOD options That would be great. But I have feeling that the cap is meant to prevent competitive VOD options from reaching critical mass.
So far, we're going to have Docsis 3.0 with the same Caps. What comes next and when? We've already had the analog shutoff.
It seems to me that Comcast's approach to competitive VOD is to have a cap. No matter, Comcast has it's own revenue generating VOD and premium channels.
Verizon is happy to sell you TV but they are not trying to channel their userbase the same way or threatened by competitive VOD services. | |
| | | | |  funchordsHelloPremium,MVM
join:2001-03-11
Yarmouth Port, MA
kudos:6
2 edits | Again, opinion cap is meant to prevent competitive VOD options said by IPPlanMan:That would be great. But I have feeling that the cap is meant to prevent competitive VOD options from reaching critical mass. To me, the question is only partially about intent. If it does prevent competitive VOD options from reaching critical mass, then such a cap should be heavily scrutinized -- intent or no.
--
Robb Topolski -= funchords.com =- District of Columbia -- KJ7RL
Evil does seek to maintain power by suppressing the truth, or by misleading the innocent. --Spock and McCoy stardate 5029.5 | |
| | | | | | IPPlanManHoly Cable Modem Batman
join:2000-09-20
Washington, DC
kudos:1
2 edits | Again, opinion cap is meant to prevent competitive VOD options Isn't it doing that already?
There are plenty of other uses for the Internet besides VOD.
The cap is creating an effective data squeeze here, exacerbated by the fact that there is no accurate meter available. | |
| | | | | | | funchordsHelloPremium,MVM
join:2001-03-11
Yarmouth Port, MA
kudos:6
1 edit | Again, opinion cap is meant to prevent competitive VOD options said by IPPlanMan:Isn't it doing that already? There are plenty of other uses for the Internet besides VOD. The cap is creating an effective data squeeze here, exacerbated by the fact that there is no accurate meter available. How would we know?
I have examples of caps and metering crushing businesses, but nobody with a 250 GB-ish soft cap, Comcast is a bit unique in this area.
The specter of the nation's largest residential ISP having a static cap year after year means that anyone who does have a 3-5 year outlook on their high-bandwidth project is not likely to fund the investment. I know that, you know that, but we don't know that it has -- in fact -- already happened.
--
Robb Topolski -= funchords.com =- District of Columbia -- KJ7RL
Evil does seek to maintain power by suppressing the truth, or by misleading the innocent. --Spock and McCoy stardate 5029.5 | |
| | | | | | | | | | | | | | | |
4 edits | Broken Funding around High Bandwidth Applications said by funchords:I have examples of caps and metering crushing businesses, but nobody with a 250 GB-ish soft cap, Comcast is a bit unique in this area. There are probably just as many, or more, examples of successful broadband businesses that grew under "invisible caps" and since.
said by funchords:The specter of the nation's largest residential ISP having a static cap year after year means that anyone who does have a 3-5 year outlook on their high-bandwidth project is not likely to fund the investment. I know that, you know that, but we don't know that it has -- in fact -- already happened. Part of the problem is funding the Usage. I think most with some basic network knowledge understand that bits cost money. More bits = more costs. Most also understand that in a Consumer flat-fee billing model, the light+medium fund the majority of heavy users and the costs wash out. The ultra-heavy are a bit different (see ad-nauseum previous discussions on this).
What few understand is that the current Internet transit/peering model does not actually fund high-bandwidth Usage properly.
[Below is mostly from a previous post in another thread]
There is a calculate shift by Content and CDNs to move their high-bandwidth growth costs to the ISPs (and in turn the Consumer). The cost of traffic growth around next gen high-bandwidth apps is something Content and CDNs, obviously, want to pay as little as possible for. Most of the time they pay an intermediate provider that doesn't actually carry the traffic any distance. If the source of traffic no longer pays their network share for the growth, or the broadband carrier that has the majority of the costs is not funded by Usage, there is a challenge. (Please read on before you call me Ed Whitacre)
Examples like ESPN360 may be the future and it is unclear if this is good or not. Also the BBC iPlayer over P2P or the BBC PR campaign to force multicast peering. This gives the BBC unlimited free access and all the end to end traffic growth costs are pushed to the broadband ISPs without incremental revenue that was previously obtained (by someone) via Content transit. While Content may not have paid that specific broadband ISP directly, with peering ratio balance requirements, this typically can wash out as a balance of trade.
The major CDNs and major Content distributors have very creative ways to cause ISPs pain and manipulate peering relationships to either gain peering (typically with smaller broadband w/ transit costs) or exploit an intermediary ISPs peering relationship with a broadband ISP. Two examples were above, and the below quoted blocks are what Content providers are stating:
Broadband ISP - Not sure I should peer with you as I have all the network costs"
quote:
We both pay for transit, so why don't we just peer
Because any transit savings for the broadband ISP are for current or existing traffic. Unlimited free (for Content) bandwidth has the potential of large end to end, unexpected, growth costs of infrastructure. These new costs no longer have a growth funding source and the costs have moved to the consumer.
Unbalanced peering traffic growth is not mutually beneficial and is an enabler which brings with it large unfunded costs.
I'm concerned about the traffic growth and while I won't offer free peering, as I have all the network costs, I will offer you the same or lower rate for transit that you pay now
quote:
If you won't peer with me, I will use your most expensive transit provider and they will make money off of you.
or
I will connect to the same router port as you and your peer and pay them small amounts unless you peer with me. (Given the costs of a port vs the full end to end costs are small, the intermediate provider can do this. This is one of the reasons the existing Internet peering environment is a mess. NOTE: Depeering is messy and not a simple solution to this)
IOW:
I will not pay you the same amount I would pay an intermediate provider even though you carry the bits end to end and they just offer 1 router hop.
All of these shifts have the potential to move the cost of high-bandwidth traffic growth from Content, to the Consumer.
You can say "I don't care, I buy bandwidth from my ISP", "Just upgrade your network", but the reality is bandwidth has been a two party payer system and the business plan for high-bandwidth apps does not fit well with the legacy Tier based transit / ISP system. | |
| | | | | | | | | funchordsHelloPremium,MVM
join:2001-03-11
Yarmouth Port, MA
kudos:6 | Re: Broken Funding around High Bandwidth Applications I have some basic networking knowledge. More bits or less bits = same cost. It's capacity (idle or not) that drives costs. The rest is just power.
The facts you cite seem to forget that the source ends must build for capacity or move data closer to the edges at their own cost. They don't just wake up one morning with a plan to jerk ISPs around, nor can they push their bits onto ISPs without some user requesting the content in the first place.
I've never seen this "incremental revenue" of which you speak. It's never existed. Someone is trying to invent it by "monetizing" my activity. I already pay a premium rate over per-hour or flat-rate dial-up service. And since Moore's law (1995 150 MHz CPU $1500 500 MB RAM 40 GB HDD; 2009 3000 MHz CPU $200 3000 MB RAM 500 GB HDD) indicates that growth ought to come for free or even at a discount, I am persuaded that there are some non-market forces at work here.
--
Robb Topolski -= funchords.com =- District of Columbia -- KJ7RL
Evil does seek to maintain power by suppressing the truth, or by misleading the innocent. --Spock and McCoy stardate 5029.5 | |
| | | | | | | | | | | | | | | | | | | |
1 edit | Re: Broken Funding around High Bandwidth Applications said by funchords:I have some basic networking knowledge. More bits or less bits = same cost. It's capacity (idle or not) that drives costs. The rest is just power. While it is true there is no incremental cost if there is available capacity. Most networks build throughout each year to keep up with capacity demand. This equals more power, as well as, more router cards, optical transponders, chassis, space, fiber, cross connects, installation costs, etc. As you stated and have stated before, this is nothing new. But it is not "just power".
What is new is the change of the two payer system for network needs to a one payer system as well as a few other interesting trends around recent broadband speed curve, transit prices, and video/high-bandwidth applications. Again, not calling exaflood, but something to watch closely and not dismiss out of hand.
said by funchords:The facts you cite seem to forget that the source ends must build for capacity or move data closer to the edges at their own cost. . Good point, but I was focused on the network side. Content and CDNs garner their incremental revenue based on usage and Content also has ad-based variable usage revenue. The more used, the more $$... In the past this also held true for their network usage, but my point is that is changing.
said by funchords:I've never seen this "incremental revenue" of which you speak. It's never existed. Perhaps you misunderstand my statement. 95%ile billing on commercial transit has been around since the 90's. The more used, the more spent. I meant the transit cost of Content vs any incremental user charges (in a two party payer system) | |
| | | | | | | | | | funchordsHelloPremium,MVM
join:2001-03-11
Yarmouth Port, MA
kudos:6 | Re: Broken Funding around High Bandwidth Applications said by yt:said by funchords:I've never seen this "incremental revenue" of which you speak. It's never existed. Perhaps you misunderstand my statement. 95%ile billing on commercial transit has been around since the 90's. The more used, the more spent. I don't think there was a point in time where
A) a content provider started pushing out data and
B) the event in "A" above resulted in a higher payment to the ISP from someone other than the ISP's customer (if even that)
Nothing has changed, here, and there's plenty of competition (and low costs) in the transit space. The rates of year/over/year growth are actually lower in the past 5 years than the decade or so before that.
Again, this being the Comcast area, it's almost not fair because Comcast is halfway through a technology upgrade that renders the whole conversation moot. But take the clock back 18-36 months ago and it really applies: there exists fiber to the home and it's wonderful. We have a legacy, leaky cable system that can only rightfully sell approximately 4 Mbps/256 Kbps connections without users draining the pooled bandwidth DOCSIS creates. Oh yeah, and we want to get into the voice business so we need to take back some bandwidth for that. Solution -- quietly kick off the big users and attack the big technologies (Sandvine RSTs) and jack up the modem speeds so we look more capable than we actually are. They did that and got busted for both. In the meantime, we actually learned the data that showed that bandwidth growth is not so bad.
In the past 18 months, more and more studios have put their stuff online for free or for cheap. This has further reduced file-sharing's uptake. They've probably also seen that improved capacity means that all those transfers actually do have an end -- the baseline amount of background traffic is probably not growing as fast as it used to be.
I'm personally hopeful that Comcast is looking ahead at bringing the bandwidth instead of holding it off. Signs are that they have no taste for this scarcity game any longer. They want people who think big things about telecom to think good things about Comcast. They don't just want to be ruler, they want to be the elected ruler. I hope that's true.
--
Robb Topolski -= funchords.com =- District of Columbia -- KJ7RL
Evil does seek to maintain power by suppressing the truth, or by misleading the innocent. --Spock and McCoy stardate 5029.5 | |
| | | | | | | | | |
4 edits | Re: Broken Funding around High Bandwidth Applications said by funchords:Nothing has changed, here, and there's plenty of competition (and low costs) in the transit space. The rates of year/over/year growth are actually lower in the past 5 years than the decade or so before that. The competitive landscape in Transit is what has changed. There are more ISPs that exploit their peering relationships to "sell" other ISPs networks at below cost and Content uses arbitrage to exacerbate this. I partially described this issue in another thread and below.
In the past, imbalance was addressed through depeering. Some "bad players" today carry enough content (imbalanced) that a depeering event would cause major disruption and, again, one of the reasons the peering environment is not working well for today's Internet.
said by funchords:Again, this being the Comcast area, it's almost not fair because Comcast is halfway through a technology upgrade that renders the whole conversation moot. You are mixing up base infrastructure, with capacity growth. One is a major forklift to set the stage for growth and the other is incremental spend to enable the capacity. An example of this is DOCSIS3 and FTTH. Both of these allow the speeds and capacity to grow, but they don't enable readily available capacity. That still comes at a cost including all the access ports, metro interconnects, metro, core and interconnect infrastructure.
said by funchords:I'm personally hopeful that Comcast is looking ahead at bringing the bandwidth instead of holding it off. Signs are that they have no taste for this scarcity game any longer. They want people who think big things about telecom to think good things about Comcast. They don't just want to be ruler, they want to be the elected ruler. I hope that's true. I agree. All signs point in that direction. Speeds continue to grow. "Bad things" have stopped. Technology is pushed forward...
It is important however to watch ALL the changes. Are the statistics trending the same as the past or very different? Understand how this is not just an ISP issue and the expectation should be that everyone ensures proper funding for bits end to end. Keep a watch out for what happens around the Internet and not just one part of it. | |
| | | | | | | | | |  jlivingoodPremium,VIP
join:2007-10-28
Philadelphia, PA
kudos:1
1 edit | What do you think will be needed in bandwidth in 5 years? said by funchords:I'm personally hopeful that Comcast is looking ahead at bringing the bandwidth instead of holding it off. Signs are that they have no taste for this scarcity game any longer. They want people who think big things about telecom to think good things about Comcast. They don't just want to be ruler, they want to be the elected ruler. I hope that's true. [personal hat on]
I think it is, personally speaking. That's what we're doing with DOCSIS 3.0. But it is always a balance... We'll throw more and more downstream and more and more upstream capacity in the network, and then the apps have to sort of catch up a bit. You can't out-drive the headlights too much or in the end the finance guys call you on it and you have trouble justifying the investment. But that's also the nice part of our network design, is that the investments can be fairly incremental and understandable.
[personal hat off]
So I'm also working on a 5-year bandwidth plan.... What d'ya think you want and need then? (seriously) 
--
JL
Comcast | |
| | | | | | | | | |
3 edits | What do you think will be needed in bandwidth in 5 years? I think it would be a good idea to map out what kinds of speed was available 5 years ago, and make a graph showing the increase. Then make a trend line, and add 15-30%
Comcast could help pull America to the top of the average internet connection speed statistics.
"The median download speed increased by only four-tenths of a megabit per second (from 1.9 mbps to 2.3 mbps), and the median upload speed barely changed (from 371 to 435 kbps). At this rate, it will take the United States more than 100 years to catch up with current Internet speeds in Japan." | |
| | | | | | | | | |  CjaicemanPremium,MVM
join:2004-10-12
Parker, CO
kudos:2
1 edit | said by jlivingood:So I'm also working on a 5-year bandwidth plan.... What d'ya think you want and need then? (seriously) More upload. I'm actually happy with the 24mbps download I get now out of my 22/5, but I would be a lot more happy with a 25/25, which is easily done on D3 once upstream bonding is rolled out (and once the CMTS's get gold certs from Cable Labs). 50/50 would great and I would jump all over that. The only thing better would be 100/100, which would be a stretch on a D3 network, but could be done w/ 4 channels in each direction, but by then there may be 8x4, so 100/100 wouldn't even stress the download side with a lot of users on.
A basic residential user would be happy I think with a 20/10.
--
Duct tape is like The Force – it has a light side and a dark side, and it binds the Universe together | |
| | | | | | | | | |
3 edits | What do you think will be needed in bandwidth in 5 years? I agree. I think all packages should be symmetrical, and in 10mbps increments all the way to 100/100 or higher. 8 channels bonding both ways is probably a good idea. Maybe try to get it down to below 50 users per node, or go FTTH. | |
| | | | | | | | | |  EGThe wings of lovePremium
join:2006-11-18
Union, NJ
kudos:9
1 edit | What do you think will be needed in bandwidth in 5 years? Where does it end... ? | |
| | | | | | | | | |
1 edit | What do you think will be needed in bandwidth in 5 years? where does what end? | |
| | | | | | | | | | EGThe wings of lovePremium
join:2006-11-18
Union, NJ
kudos:9
2 edits | What do you think will be needed in bandwidth in 5 years? Keeping up with the *Joneses* ??  | |
| | | | | | | | | |
1 edit | What do you think will be needed in bandwidth in 5 years? Until you hit 100 Mb/s? | |
|
| | | | | | | | | funchordsHelloPremium,MVM
join:2001-03-11
Yarmouth Port, MA
kudos:6 | said by jlivingood:So I'm also working on a 5-year bandwidth plan.... What d'ya think you want and need then? (seriously) In 2014, assuming status quo capping
$40/mo, 50 Mbps/10 Mbps, up to 2 TB usage
--
Robb Topolski -= funchords.com =- District of Columbia -- KJ7RL
Evil does seek to maintain power by suppressing the truth, or by misleading the innocent. --Spock and McCoy stardate 5029.5 | |
| | | | | | | | | | IPPlanManHoly Cable Modem Batman
join:2000-09-20
Washington, DC
kudos:1
1 edit | Re: What do you think will be needed in bandwidth in 5 years? I'll bet that by that time, 2TB will feel like 250GB... Seriously. Technology does not have a linear growth pattern. It grows exponentially. | |
| | | | | | | | | |
1 edit | Re: What do you think will be needed in bandwidth in 5 years? how about like this:
"15% of the total amount of bandwidth you could use in a month"
you could increase the percentage over time
i.e.
50/10 Mb/s tier would get: 2.5 TB
22/5 Mb/s tier would get: 1 TB
16/2 Mb/s tier would get: .75 TB
12/2 Mb/s tier would get: .5 TB
8/2 Mb/s tier would get: .35 TB
6/1 Mb/s tier would get: .25 TB | |
| | | | | | | | | | | | | | | | | | | |
2 edits | If metered, What cost would you pay? What cost per GB or 10GB would you pay? | |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
See 8 replies to this post |
| | | | | | | | |  JohnInSJPremium
join:2003-09-22
San Jose, CA
·PHONE POWER
 ·Comcast
| Re: If metered, What cost would you pay? said by nate1234:What cost per GB or 10GB would you pay? Let's look at Amazon Ec2 transport costs (why not?)
from »aws.amazon.com/ec2/#pricing
Internet Data Transfer
The pricing below is based on data transferred "in" and "out" of Amazon EC2.
Data Transfer In
All Data Transfer $0.10 per GB
Data Transfer Out
First 10 TB per Month $0.17 per GB
Next 40 TB per Month $0.13 per GB
Next 100TB per Month $0.11 per GB
Over 150 TB per Month $0.10 per GB
So, you're looking at 10 cents per GB to 17 cents per GB. 250GB is $25-$42.50 depending on the in/out mix - call it $35 on average.
That's probably a bit more then the carry cost for Amazon (ie, they make a buck )
Using a hosting service like Godaddy, you'll get 300GB/mo for your $5/mo web site. That's pretty cheap. Want cheaper? inmotionhosting.com gives you 1TB/mo on their cheap personal hosting plan.
This show you that the cost is associated more with the customer side then the backhaul - apparently backhaul is nearly free.
Funny Aside... »www.isp-planet.com/business/pricing3a.html - in 1999 using a T1 full out would cost you $4/GB! Your 250GB allotment from comcast would be the same as using a full T1 for 35 days, or $1000.
--
My place : »www.schettino.us | |
| | | | | | | | | | | | Re: If metered, What cost would you pay? Inmotionhosting.com = $7/mo for unlimited bandwidth and disk space! | |
| | | | | | | | | | JohnInSJPremium
join:2003-09-22
San Jose, CA Reviews:
·PHONE POWER
·Comcast
| Re: If metered, What cost would you pay? said by nate1234:Inmotionhosting.com = $7/mo for unlimited bandwidth and disk space! Nice, but bandwidth is only free after you get out of a neighborhood. *all* (essentially) of comcasts costs are NOT backhaul.
--
My place : »www.schettino.us | |
|
| | | | | | | | | | | You should make a web page or sticky'd topic that contains all of those IETF documents easily available, so we don't have to look through sub-threads.
It would be great  | |
|
| | | | | | | | |
2 edits | Re: Broken Funding around High Bandwidth Applications said by funchords:I don't think there was a point in time where A) a content provider started pushing out data and B) the event in "A" above resulted in a higher payment to the ISP from someone other than the ISP's customer (if even that) It took me a while to understand this and let me repeat it to make sure I got it right. You are saying that there has never been a point in time where a content provider started sending traffic at increased levels that resulted in a higher payment to an ISP other than that customers ISP or an unrelated ISP from their respective customer.
If that is accurate, I would say you are correct. And this is made possible by trade balance between the respective ISPs. IOW I agree not to send more traffic to you than you send to me and therefor will not exploit our peering relationship. This breaks down a bit with P2P delivery or ISP C like tactics of selling other ISPs capacity at reduced rates.
While it sounds like you get the ISP to ISP dispute, this same concept holds true with Content suppliers. Bandwidth is not free and it is up to both Users AND Content to each pay their portion of the end to end capacity. If one party does things that shift those costs to the other, I would rather see the shift move from Users to Content and find ways to subsidize network growth via other means.
Current State: As usage increases, Users pay the same, ISPs may have more costs (depending on technology/capacity/etc), Content makes more money. | |
| | | | | | | | | | IPPlanManHoly Cable Modem Batman
join:2000-09-20
Washington, DC
kudos:1
1 edit | Metered billing system, discussed many times A metered billing system addresses this problem better than anything else. Why should someone who uses 2-3 GB a month be charged the same as someone who uses 90-100x that?
Power companies don't do that. If someone runs their A/C all the time, they pay for it. We all don't pay the same electric bill. Imagine the outrage if we did.
--
"We're going to start at one end of (Fallujah), and we're not going to stop until we get to the other. If there's anybody left when that happens, we're going to turn around and we're going to go back and finish it."
Lt. Col. Pete Newell: 1st Inf. US Army | |
| | | | | | | | | |
1 edit | Metered billing system, discussed many times I dont want metered billing... | |
| | | | | | | | | | IPPlanManHoly Cable Modem Batman
join:2000-09-20
Washington, DC
kudos:1
2 edits | Metered billing system, discussed many times Why not? Phone gas and electric are metered. Are you unhappy with that? If so, then tell them you don't want it.
Use more, pay more. | |
| | | | | | | | | | |
| | | | | | | | | espaethDigital PlumberPremium,MVM
join:2001-04-21
Minneapolis, MN
kudos:2 Reviews:
·Vitelity VOIP
| said by funchords:I have some basic networking knowledge. More bits or less bits = same cost. It's capacity (idle or not) that drives costs. The rest is just power. You have to understand that with the topology under which networks are constructed, links are sized to meet demand so utilization does have a direct impact on costs.
Take a data center access model. In a Cisco Catalyst 6509 chassis you get 9 slots to work with, 2 slots are occupied by supervisors (the 'brains' of the switch), and the other 7 slots can be used for line cards. The backplane capacity is 40gbps per slot. Say you are building out a data center of servers all attaching in with GigE links, all plugging into WS-X6748-GE-TX line cards which provide 48 ports of attachment with a 40Gig slot attachment (1.2:1 oversubscribed). That's great for attachment, but the network is only useful if you can go somewhere, so you install a WS-X6704 4 port 10GigE module to attach that switch into an upstream distribution switch, which attaches to the core. So you end up with 6 x 48 (288) edge ports of attachment and 4 x 10GigE ports of upstream attachment.
This works great until some application comes along that drastically changes the traffic you see out of the infrastructure -- say something like database replication to diverse locations for site failure survivability that drives your demand to 60Gbps leaving the access switch. Now you're stuck with a problem -- you haven't increased the link speed on any server, but now utilization causes a change in architecture. You need to setup a new switch chassis (complete with expensive supervisors and power supplies) with yet another 10GigE line card for uplinks and 48 port line cards for server attachment, you need to migrate servers from the existing switch over to it so that you can reduce some of the bandwidth demand and also free up slot capacity to install another 10GigE line card. This additional capacity often drives expansion of the hardware at the distribution layer, which may need to be broken out into even more switches, which also pushes expansion upwards to the core layer as well.
How does this relate to the cable plant? Node splits for capacity require additional CMTS ports, which drives the need for more CMTS chassis, which drives the need for distribution upgrades, etc.
Oversubscription is what makes modern networks work. I'm not just talking data networks, because voice networks use the same model. There are events such as natural disasters which cause phone networks to become overloaded; they could be sized such that such a limit could not be reached, but most people would never be able to afford phones if that were the case.
said by funchords:And since Moore's law (1995 150 MHz CPU $1500 500 MB RAM 40 GB HDD; 2009 3000 MHz CPU $200 3000 MB RAM 500 GB HDD) indicates that growth ought to come for free or even at a discount, I am persuaded that there are some non-market forces at work here. Moore's law applies to manufacturing densities, such as the quantity of transistors per chip or the density of storage on a magnetic platter. It applies to items where improvements in manufacturing process can be used to produce gains in performance or storage density.
I/O rates aren't governed by manufacturing process -- they're governed by the coding methods used to deliver signals, the physical media used for the communication, and the laws of physics. Hard drives are a great example, because while the increase in quantity follows Moore's law, the increase in sustained transfer rates from the drives or reduction in seek times do not.
The overall costs of providing network expansion are not going down. The CMTS ports are likely dropping a bit in price, and there are some gains to be had in head-end distribution technology improvements. Those costs tend to be the smallest dollar figure in any expansion anyway. The cost of labor continues to be more expensive, copper material costs have gone up, fiber material costs have gone up, and energy costs have continued to rise. (Energy is a double hit, because hardware that sucks down power must be cooled) A node split that requires the trenching of new fiber and re-engineering of the gain structure on the RF cable plant is more expensive today than it was even 5 years ago. That's why I believe Comcast is focusing so much on Project Cavalry, because money spent on reclaiming frequency space on the existing plant delivers far more value than simply throwing money at node splits. | |
| | | | | | | | | |
See 15 replies to this post |
| | | | | | | | funchordsHelloPremium,MVM
join:2001-03-11
Yarmouth Port, MA
kudos:6 | said by yt:Broadband ISP - Not sure I should peer with you as I have all the network costs" quote:
We both pay for transit, so why don't we just peer
Because any transit savings for the broadband ISP are for current or existing traffic. Unlimited free (for Content) bandwidth has the potential of large end to end, unexpected, growth costs of infrastructure. These new costs no longer have a growth funding source and the costs have moved to the consumer. Unbalanced peering traffic growth is not mutually beneficial and is an enabler which brings with it large unfunded costs. We're talking about ISPs here who get money from their consumers for X Mbps up and Y Mbps down. These consumers request all the data coming in the download direction and generate all the traffic in the upload direction. As long as they work within their budget of X Mbps up and Y Mbps down, then their unlimited ISP ought to leave them alone or set a limit Z (booo but okay). If they want more than X and Y and Z, then they can pay comparatively more than those that want less.
Peering gets the middleman and his bills out of the way. ISPs ought to do it regardless of any balance, because balance just doesn't matter. It's an access provider -- it's generally going to have more coming down than going up, but the source of all the activity is that provider's customers.
--
Robb Topolski -= funchords.com =- District of Columbia -- KJ7RL
Evil does seek to maintain power by suppressing the truth, or by misleading the innocent. --Spock and McCoy stardate 5029.5 | |
| | | | | | | | | | | | Re: Broken Funding around High Bandwidth Applications said by funchords:Peering gets the middleman and his bills out of the way. ISPs ought to do it regardless of any balance, because balance just doesn't matter. That is a common statement by most Content/CDN folks. While it may be true for small ISPs that outsource most of their network costs to transit ISPs, it is not true for larger ISPs that built and support national/international infrastructures. Once you dig a bit deeper, most understand that network growth has cost as much as system growth has costs. Content folks do get this. They just don't want to talk about it.
The Internet peering world was original built on balance of trade. Ratios have business merit as the network has a cost to carry. Always had, always will. The difference is ISPs have evolved and some have capitalized on ratio imbalance to "dump" traffic at the expense of the receiver. There have been some very public de-peering events related to this. | |
| | | | | | | | | | funchordsHelloPremium,MVM
join:2001-03-11
Yarmouth Port, MA
kudos:6 | Re: Broken Funding around High Bandwidth Applications said by yt:The Internet peering world was original built on balance of trade. Ratios have business merit as the network has a cost to carry. Always had, always will. The difference is ISPs have evolved and some have capitalized on ratio imbalance to "dump" traffic at the expense of the receiver. There have been some very public de-peering events related to this. I kinda get that when it's two transit providers who are fighting. But it never makes sense to me when one end is an ISP responsible for generating both the outgoing request and the incoming result.
--
Robb Topolski -= funchords.com =- District of Columbia -- KJ7RL
Evil does seek to maintain power by suppressing the truth, or by misleading the innocent. --Spock and McCoy stardate 5029.5 | |
| | | | | | | | | |
2 edits | Re: Broken Funding around High Bandwidth Applications said by funchords:I kinda get that when it's two transit providers who are fighting. But it never makes sense to me when one end is an ISP responsible for generating both the outgoing request and the incoming result. There is still an end to end cost of carrying bits. Instead of $/Mb think $/Mb/Mile. To further complicate this, understand that some miles cost more than other miles (e.g. last mile).
If I am ISP A and I carry content 100 Miles to my peer ISP B who then carries it 500 miles… AND I am the one that is paid for it as the Transit provider for that content, then I make out better. However, that same ISP B has situations where they carry it 100 miles and I carry it 500 miles (this is how hot-potato routing works). If both ISPs have a fairly balanced ratio of in vs out traffic then both have equal cost burden and the system is fair. This is how it use to work.
Now we have ISP C that likes the 100 mile part and sells bandwidth cheaper (because they only have to carry it 100 miles). They undercut the cost and take the other ISPs business. Traffic goes imbalanced as one ISP is exploiting the others. ISP C is now so big that depeering is a major disruptor.
Make sense so far?
Now to the next phase. Content goes to ISP C and says I will connect to the same router port as you are connected to ISP A and ISP B and your costs go from 100 miles to 10 meters. Wow… sounds great for ISP C and Content. ISP A and B however have to carry that traffic 500+ miles and lose revenue to ISP C unless they sell for possibly below cost to carry. Traffic goes even more imbalanced and balance of cost/revenue no longer exists. The ISPs carrying the traffic the shortest distance, sell the most for the least. Good for Content. Good for ISP C. Not so good for ISP A and B.
Content then goes to ISP A and B and says peer with me. | |
|
| | | | | | | | funchordsHelloPremium,MVM
join:2001-03-11
Yarmouth Port, MA
kudos:6 | said by yt:I'm concerned about the traffic growth and while I won't offer free peering, as I have all the network costs, I will offer you the same or lower rate for transit that you pay now quote:
If you won't peer with me, I will use your most expensive transit provider and they will make money off of you.
or
I will connect to the same router port as you and your peer and pay them small amounts unless you peer with me. (Given the costs of a port vs the full end to end costs are small, the intermediate provider can do this. This is one of the reasons the existing Internet peering environment is a mess. NOTE: Depeering is messy and not a simple solution to this)
IOW:
I will not pay you the same amount I would pay an intermediate provider even though you carry the bits end to end and they just offer 1 router hop.
All of these shifts have the potential to move the cost of high-bandwidth traffic growth from Content, to the Consumer. "The amount of traffic being routed through my most expensive transit provider is going to be proportionately small. The net effect is that your customers are going to become frustrated with your site because it will be slow. If they call me, I'll blame it on your transit provider because they're the ones with the unattractive connectivity costs."
Who is jerking who around? There's nothing non-neutral about that (before anyone says there is).
You can say "I don't care, I buy bandwidth from my ISP", "Just upgrade your network", but the reality is bandwidth has been a two party payer system and the business plan for high-bandwidth apps does not fit well with the legacy Tier based transit / ISP system. If the speed and capacity of our network grew in step with technological advances, I might be more sympathetic. But the fact is that around the world, high-speed users are getting more and paying less for it. Our claim to fame on the worldwide stage is the highest bills. We're even going backward -- from unlimited to limited.
If you're tired of the chant, "Just upgrade your network," then tough! It works, and it's what we're going to do! I've seen this play out time and time again -- 1988, 1995, 2000, 2003, and 2007 where the demand curve hits the technology step-function due to growth, images, music, and now high-res video. All the bitching, filtering, caching, clustering, pushing, pulling all eventually leads to a point where the network grows and the pain goes away for a while and we have a few years of wondering "what will ever fill these pipes now? was this a big waste?"
--
Robb Topolski -= funchords.com =- District of Columbia -- KJ7RL
Evil does seek to maintain power by suppressing the truth, or by misleading the innocent. --Spock and McCoy stardate 5029.5 | |
| | | | | | | | | | | | | | | | | | | | | | Re: Broken Funding around High Bandwidth Applications said by funchords:I've seen this play out time and time again -- 1988, 1995, 2000, 2003, and 2007 where the demand curve hits the technology step-function due to growth, images, music, and now high-res video. Again, my point was not "can we keep up". It was to point out that the Funding model of high-bandwidth growth is different as ISPs have evolved, CDNs grew, and the peering/transit business model is about 20 years old. Combine that with speeds have never grown this fast, transit prices have never dropped this fast for Content and it is something interesting to experience.
This is a different perspective of the issue. While it is good to judge the future based on our past experiences, we should watch for the variables in the equation. | |
|
| | | | |
1 edit | said by funchords: If it does prevent competitive VOD options from reaching critical mass, then such a cap should be heavily scrutinized -- intent or no. It already prevented it for me.
--
Obama '08. Will help resolve the terrible broadband issues we have that put us so far behind other countries. | |
|
| | | | espaethDigital PlumberPremium,MVM
join:2001-04-21
Minneapolis, MN
kudos:2 Reviews:
·Vitelity VOIP
1 edit | DOCSIS 3.0 has the same cap, so far said by IPPlanMan:So far, we're going to have Docsis 3.0 with the same Caps. Not all areas have been upgraded to DOCSIS 3.
Greatest common divisor is the easiest approach when setting the policy, considering 99.99+% of the subscriber base won't be affected by it.
A 38mbps DOCSIS channel cannot move more than 12312GB/mo. When you factor out overhead and CDV bandwidth, the HSI share is still divided out across approximately 250 users per channel today. If you do the math, it's not possible to support every user hitting 250GB.
Over time that will improve, but you have to keep in mind that network expansion is done over years, not months. | |
| | | | | |
See 12 replies to this post |
|
| |
|
