dslreports logo
site
 
    All Forums Hot Topics Gallery
spc

spacer




how-to block ads


Search Topic:
uniqs
14
share rss forum feed


espaeth
Digital Plumber
Premium,MVM
join:2001-04-21
Minneapolis, MN
kudos:2
reply to Lazlow

Re: Woo Hoo! Free capacity!

said by Lazlow:

The part you are forgetting (and ISPs are trying to hide) is that the cost per/unit has been and will continue to drop dramatically.
Ahh... but *why* and *where* has the cost per unit dropped? At the core, companies have been able to take leased fiber and swap out GigE optics for 10GigE optics or even xWDM optics to get N*10GigE on the same physical fiber pair.

Upgrades at the core are easy because: the technology is there to provide 1000+% increase in capacity over GigE on the same fiber path, the interfaces are available for existing network hardware already used in core positions, and the core represents a handful of physical connections in relation to all of the ports at the edge.

Growing bandwidth at the edge isn't so simple. Hardware is cheap and purpose built. You want to upgrade to DOCSIS 3.0, you need to swap out every single cable modem on the plant to get the true full benefit. You want to upgrade to VDSL? You need to deploy more remotes closer to the subscribers at massive costs *AND* swap out every existing customer DSL modem with a VDSL capable model. You want to upgrade FiOS BPON to GPON, you have to replace every ONT on the network segment you upgrade to be GPON. Even then, you still top out at about a 300% increase in capacity over what you have today at ridiculously high costs.

The growth model at the core and the edge is highly disproportional. At the edge you spend a ton of cash to get minimal gains; at the core you can get massive gains with minimal cash investment.

Lazlow

join:2006-08-07
Saint Louis, MO
"You want to upgrade to DOCSIS 3.0, you need to swap out every single cable modem on the plant to get the true full benefit."

espaeth

That is patently false. Docsis 3.0 is backward compatible to previous versions. The only cable modems that need to be replaced are those of the high speed users. You bond their channels off the channels(on different channels than) used by the earlier Docsis versions. This gives the high speed users (presumably high capacity as well) a clear path and it drops them off the capacity load on the channels being used by the older Docsis versions (more bandwidth for the Docsis 2.0 users to share). Yes, only those users that have the Docsis 3.0 modems will be able to get the speeds above 30meg(I think that is the limit for 2.0). But most of the ISPs are claiming that only a small percentage of their users are soaking up the majority of the bandwidth. Presumably these(high capacity) user are usually the high speed users. Even if all the above was not true, every single piece of equipment you mentioned is cheaper today than it was a year ago and a year from now it will be even cheaper. So as time goes on the price/unit drops regardless.


Dogfather
Premium
join:2007-12-26
Laguna Hills, CA

1 edit
And given the prices that cable operators like Comcast charge for their pre-DOCSIS 3 tiers ($150/mo) I think they can afford to kick down a modem.


espaeth
Digital Plumber
Premium,MVM
join:2001-04-21
Minneapolis, MN
kudos:2
reply to Lazlow
said by Lazlow:

"You want to upgrade to DOCSIS 3.0, you need to swap out every single cable modem on the plant to get the true full benefit."

espaeth

That is patently false. Docsis 3.0 is backward compatible to previous versions. The only cable modems that need to be replaced are those of the high speed users.
My statement is still accurate. Even if you build out DOCSIS 3.0 in 4 channel clusters and balance all of the existing D1.1/2.0 users across those channels, it is still easy to saturate a single channel of the bundle if your heavy users end up on the same channel.

Take a situation where 5 users with 8mbps downstream connections all try to max out their downstream capacity. On D1.1/2.0 platforms with single channel attachment, if they all land on the same channel they will exceed the 38mbps channel capacity and congestion will result.

The full benefit of D3.0 would be if those users all had access to the 152mbps (38x4) shared bandwidth pool. In that case it would be 40mbps out of a total channel capacity of 152mbps so there would be plenty of headroom left over.

There are definitely gains to be had just by adding the additional downstream channels, but the real full benefit doesn't come until a significant portion of your user base has converted.

said by Lazlow:

But most of the ISPs are claiming that only a small percentage of their users are soaking up the majority of the bandwidth. Presumably these(high capacity) user are usually the high speed users.
Not necessarily. I illustrated my point above with just (5) 8mbps users on a segment. Look at all the people in the forums talking about running P2P on the entry-level cable tiers.

said by Lazlow:

Even if all the above was not true, every single piece of equipment you mentioned is cheaper today than it was a year ago and a year from now it will be even cheaper. So as time goes on the price/unit drops regardless.
DOCSIS 3.0 modems are significantly more expensive than DOCSIS 2.0 modems. They have multiple radio tuners which adds costs to the manufacturing. The DOCSIS 2.0 stuff will continue to decline in price as demand diminishes.

The problem is that upgrades at the edge generally aren't about simply adding more, it's about forklift upgrades. DOCSIS 3.0 at the head-end required the complete replacement of DOCSIS 2.0 CMTS hardware with D3.0. To tap in the the capacity at the home you need a new modem. Growth at the edge is a matter of replacement, not expansion.


espaeth
Digital Plumber
Premium,MVM
join:2001-04-21
Minneapolis, MN
kudos:2
reply to Dogfather
said by Dogfather:

And given the prices that cable operators like Comcast charge for their pre-DOCSIS 3 tiers ($150/mo) I think they can afford to kick down a modem.
The entry price for pre-DOCSIS 3 service for Comcast is $62.95/mo for the 22/5 tier. (the obvious qualification is that your area needs to be upgraded to get that)

Lazlow

join:2006-08-07
Saint Louis, MO
reply to espaeth
On the Docsis channel thing: IF they would stop advancing speeds for a while and increase load capacity (until they upgrade the system to handle the load) it will be a non issue. I am assuming that you did not intend to share the same channels between Docsis 3 and pre-Docis3 in your example(that would essentially defeat the purpose). Getting those high capacity users over to Docis 3 modems would (of course) be a priority. Getting those users off the pre-3.0 channels should open a significant enough amount of bandwidth to handle the next year or two(?). The higher end users will be motivated to switch over to 3.0 modems so that they see less congestion (self curing problem). The rest will probably see an improvement over the situation that they are in today and will slowly migrate over to the 3.0 modems (probably about the time 8 channel is coming in).

You cannot compare the prices of Docsis 2.0 modems to 3.0 modems. You need to compare apples to apples. What was the price of a 3.0 modem a year ago (still prototype?) and what is it today? Compare the price of a D3 cmts today and a D2 cmts when it was introduced. While I do not know for certain I am willing to bet that the price for the D3 CMTS in real dollars is significantly less than(but possibly the same due to the increased number of transmitters/tuners) the D2 units.


espaeth
Digital Plumber
Premium,MVM
join:2001-04-21
Minneapolis, MN
kudos:2
said by Lazlow:

On the Docsis channel thing: IF they would stop advancing speeds for a while and increase load capacity (until they upgrade the system to handle the load) it will be a non issue.
It's somewhat counter intuitive, but in a statistically-muxed network increasing access speeds actually tends to improve performance across the board. The reason for that is simple: the less time it takes me to get the content I want, the faster I'm off the network, and the more "idle" time that will be available to the other users on the network.

Applications like P2P that assume that all "idle" network capacity belongs to you obviously kill any stat-mux gains of finite duration transfers.

said by Lazlow:

I am assuming that you did not intend to share the same channels between Docsis 3 and pre-Docis3 in your example(that would essentially defeat the purpose).
You don't have the luxury of *not* sharing between D3 / non-D3 modems. DOCSIS 1.1 / 2.0 modems will attach to the first channel they successfully tune to; DOCSIS 2.0 at least has the ability to be sent a channel change command so they can be distributed across the channels. Open channel space is a precious commodity for MSOs, they had to fight to make room for the 3-4 channels of DOCSIS 3 they are deploying now.

Lazlow

join:2006-08-07
Saint Louis, MO
While I agree with the first point to a certain extent, I think it is a matter of degree. A lot of systems went from 5meg (top service) to 16meg without a similar increase in capacity increase. Jumping from 5-10 this way was "in range" but jumping on up to 16 just tipped the cart over.

With the number of analog channels that have been removed from service on most systems they could easily isolate the Docsis 3 from the pre3 channels. We both know that you can define the list of channels that the modems can select from and that you can use separate lists for 3/pre3. But they want that recovered bandwidth for HD only. Again it goes back to greed.


espaeth
Digital Plumber
Premium,MVM
join:2001-04-21
Minneapolis, MN
kudos:2
said by Lazlow:

While I agree with the first point to a certain extent, I think it is a matter of degree. A lot of systems went from 5meg (top service) to 16meg without a similar increase in capacity increase. Jumping from 5-10 this way was "in range" but jumping on up to 16 just tipped the cart over.
To be fair, when Comcast started deploying Blast! on the east coast they put a lot of money into node splits to reduce subscriber density per DOCSIS channel. It's not like they just rolled out new profiles without making any changes to the infrastructure.

said by Lazlow:

With the number of analog channels that have been removed from service on most systems they could easily isolate the Docsis 3 from the pre3 channels. We both know that you can define the list of channels that the modems can select from and that you can use separate lists for 3/pre3. But they want that recovered bandwidth for HD only. Again it goes back to greed.
In order to do the most effective upgrade you do a one-for-one replacement of DOCSIS 1.1/2.0 CMTS hardware with DOCSIS 3.0. The old equipment doesn't stick around, if for no other reasons than the diagnostic capabilities of the D3.0 line cards are so much better than earlier revisions. (although the real reason tends to be space, power, fiber constraints) Since DOCSIS 3.0 requires new modems in the field, to not put your existing clients on the same channels would be pointlessly wasteful. The DOCSIS 3 modems will balance around the available channels, so if one happens to become saturated the traffic will simply be diverted to the other links -- it can route around single channel congestion.

As far as them "keeping the channels for HD" -- let's be honest here, it's not for "them," it's for the customers that call them up every day asking for more HD channels. Not to mention that a vast many customers are actually willing to pay more for more HD content. Reserving capacity for services that are profitable is what smart businesses do.

Lazlow

join:2006-08-07
Saint Louis, MO

1 edit
"Since DOCSIS 3.0 requires new modems in the field,"

NO, it does not. It is backward compatible with older versions of Docsis. A Docsis 2.0 modem will function just fine on a Docsis 3.0 CMTS. No, the older modems will not be able to take advantage of the newer 3.0 features. Which gets us right back where we started from. Where I said the "power users" (or whatever term you like) will be the first to upgrade to the 3.0 standard (to gets those features) and the rest will upgrade over time (just like the migration from 1.0 to 2.0). You should still be able to split those channels (3.0 from pre3.0) so that you can increase the capacity of both the 3.0 and the pre3.0. In the end that is what will occur when they switch to 8 channel anyway.

IF you do not think that virtually everything they do is in their best interest you really need to step back. As we have seen recently (bailout anyone?) there is a huge difference between "smart" business and honest business, and you are right I think they are running the same type "smart" business as the mortgage people. You can make an honest profit without a lot of the things that go on. When they offer people (current customers not new ones) 6 month or a year special deals do you really think that they are taking a loss on those deals? No, they are just dropping their profit margin. In a lot of markets 5meg service is given on special for 6months (to existing costomers) for $15/month while the standard price is $60 (stand alone price on both).

As far as Comcast Blast deal; in a lot of markets they did just drop it in without making major changes until AFTER the congestion showed up(just check the old threads). But yes, they did split a lot of those nodes to ease the problem. Which just shows that they are capable of handling marked increases in bandwidth without resorting to caps.

In the ISPs defense, the FCC screwed the pooch on this transition. They should have made everything switch at once(free up all the old analog bandwidth). But they should have made the ISPs keep the same channel line up, only broadcasting on clear Qam(for basic). The $40 digital converters should have had a qam tuner in them as well. This would have eliminated the company owned set top boxes(that we fought so hard to get rid of the first time around) and still kept a decent basic cable lineup(in digital). To go along with that they should have made no distinction between digital/analog as far as the broadcast stations were concerned(same rebroadcast rules should apply to digital as analog).


espaeth
Digital Plumber
Premium,MVM
join:2001-04-21
Minneapolis, MN
kudos:2

3 edits
said by Lazlow:

"Since DOCSIS 3.0 requires new modems in the field,"

NO, it does not. It is backward compatible with older versions of Docsis.
Quit dancing on that point, you know what I mean. You're not going to get DOCSIS 3.0 speed and capacity increases without DOCSIS 3.0 endpoints.

In relation to my earlier statement, since there will be a scarce few DOCSIS 3.0 modems out on the network in the near term it makes poor sense to set aside multiple DOCSIS channels just for a handful of D3.0 customers. It makes better strategic sense to spread your existing DOCSIS 1.1 / 2.0 clients across those channels and get some use out of them.

said by Lazlow:

IF you do not think that virtually everything they do is in their best interest you really need to step back. As we have seen recently (bailout anyone?) there is a huge difference between "smart" business and honest business, and you are right I think they are running the same type "smart" business as the mortgage people.
The mortgage people couldn't do it alone. Do you honestly think not a single person who suddenly qualified for a loan even though they never qualified before and got setup with sub-prime rates didn't think something was up? The mortgage situation has two guilty parties, and while the actual mortgage brokers should have played a better "parent" role and stopped people from making bad financial choices, it still stands that we got here because a vast many people made horrible financial choices.

Residential customers have an artificially low perception of what it actually costs to operate a network infrastructure because of the "unlimited" connection pricing that is engineered such that the average subscriber has no chance of hitting anything close to "unlimited." Customers are being sold timeshares, but they think they are buying their own permanent residence. We're in this mess now because even though people keep saying there's a huge demand for bandwidth, the most popular service tier remains the absolute cheapest service that the broadband providers offer. Customers want Max(Bandwidth) for Min(Price) -- something has to give.

said by Lazlow:

You can make an honest profit without a lot of the things that go on. When they offer people (current customers not new ones) 6 month or a year special deals do you really think that they are taking a loss on those deals?
Of course they take a loss on those sales. They want to introduce you to a product at a low introductory rate so you will stay a customer going forward. This sales tactic is used universally from street drug dealers offering a free "taste" to selling game station consoles for less than the hardware manufacturing costs with the hopes of making it up in game sales after the fact. It's called a loss leader for a reason.

Lazlow

join:2006-08-07
Saint Louis, MO

2 edits
Quit trying to insinuate that the only gains that can occur on the 3.0 systems are when all the modems are 3.0. Will the maximum gain be if all the modems are 3.0? YES, but if you can move just the top 10% of the bandwidth consumption clients over to 3.0(at least if you trust the ISPs earlier statements), most of the congestion will be a non-issue(assuming you split the channels) and without resorting to caps. Bandwidth consumption will continue to increase, as it increases more and more people will switch to 3.0 modems. By the time the vast majority has switched over to 3.0, the bandwidth consumption/speed requirements will probably be high enough to require 8 channel Docsis 3.0. So at that time the 4 channels that had been dedicated to pre3.0 use can be reintegrated to the 3.0 8 channel systems.

edit about your edit. I assume you meant scarce few? That is why the ISPs should push the 3.0 modems on those customers that are using the most bandwidth.

further edit: rather than overage charges or banning a customer just require them to move to a 3.0 modem.