dslreports logo
 
    All Forums Hot Topics Gallery
spc
uniqs
64
Skippy25
join:2000-09-13
Hazelwood, MO

Skippy25 to Crookshanks

Member

to Crookshanks

Re: Yeah, let's just ignore the access charges

I can dispute your use of the word "greater" as a person that has an average bitrate of 5mbit/s has a very marginal cost over someone that uses 100kbits/s per second and that is a fact.
tanzam75
join:2012-07-19

tanzam75

Member

said by Skippy25:

I can dispute your use of the word "greater" as a person that has an average bitrate of 5mbit/s has a very marginal cost over someone that uses 100kbits/s per second and that is a fact.

The additional cost of the heavy user is only negligible while the node has extra capacity.

As soon as the node hits capacity, the next bit costs thousands of dollars, for the additional equipment to split the node. And that's assuming you've got unlit fiber available in your existing plant.

As soon as you run out of fiber to do logical node splits, the next bit costs tens of thousands of dollars. You have no choice but to run new fiber to the new nodes.

Now, who pays for it? Do you make grandma pay the same amount as the heavy downloader? That's unfair, because grandma didn't max out the node. Wouldn't it be fair to impose a surcharge on the heavy downloaders, so that they end up paying all of the costs of the additional infrastructure?

You can argue that the caps are too low, or that the price of the surcharges is too high, or that there's no off-peak free-use period. Indeed, the cable industry would have a much easier time justifying the caps if they were to have a separate -- and much higher -- cap for off-peak usage.

But you cannot just blindly argue against caps, as so many do on the DSLReports forums, unless you deliberately ignore the fundamental economics of Internet service. "Bits are almost free!" Yes, but there's an asterisk -- they're not free when the pipes get congested.

morbo
Complete Your Transaction
join:2002-01-22
00000

morbo

Member

Yes, it costs money to upgrade a node. However, unless the network is completely mismanaged to the point of incompetence, a single user using above average resources cannot be attributed to the cost. That's like saying that the 3 lane interstate highway is congested for 5 hours a day, and at 5 hours and 1 minute of congestion per day the next driver is responsible for adding an additional lane to the highway (millions and millions of dollars). It's ridiculous.
tanzam75
join:2012-07-19

tanzam75

Member

It certainly is not ridiculous.

Indeed, fair allocation of fixed costs is precisely the idea behind toll lanes on otherwise-free highways. Because it's the peak-period drivers who are forcing the road to be expanded, it would be fair for them to pay 100% of the costs. In contrast, the off-peak drivers could've gotten by without the extra lanes, so they should pay 0% of the costs. For example, just two months ago, Los Angeles opened High Occupancy/Toll lanes on the I-110.

It's also becoming common for bridge and tunnel expansions to be paid for through time-of-day pricing. If you use it at midnight, you pay one rate, because you could've gotten by just fine on the old two-lane bridge. If you use it during rush hour, then you pay a much higher rate, because you're one of the commuters that forced the government to spend hundreds of millions of dollars on a new bridge.

That "millions and millions of dollars" argument is an illogical strawman. Nobody's asking one person to pay all of the money up-front. The fees are paid a little bit at a time, collectively, by all the people who forced the upgrade. Just like bandwidth cap overages are paid collectively, tens of dollars at a time, by all of the people exceeding the cap.

morbo
Complete Your Transaction
join:2002-01-22
00000

morbo

Member

You are not consistent with your own flawed analogy so this discussion is going no where.

With cable admitting that caps are not about congestion, your point is completely debunked. It's also not about fairness. Until grandma is paying $5 a month for checking her email and the torrent users pay $1000 a month for their heavily used connection, this isn't about fairness. It's about cable companies wanting to double dip for services that are already provided and paid for.
tanzam75
join:2012-07-19

3 edits

tanzam75

Member

Yes, it's unfair that grandma doesn't get to pay $5 for cable Internet, while the bandwidth hog pays $1000. But surely it's fairer to charge grandma $50 and the bandwidth hog $100, rather than charging everyone $60.

The problem is that bits are not like molecules -- they're free to transport, until you have to do a node split. So do you price them like they're free, with a fixed infrastructure charge? Or do you price them like they're expensive, with a metered charge? Why not a combination of both, to account for their hybrid nature?

To point to the lack of a $5 grandma rate as a reason not to charge overage fees is illogical. You think it's unfair not to charge pure usage-based pricing, so you instead advocate a flat rate in which it's even less fair?

I don't understand this "gotcha" mentality when it comes to the guy's comments on congestion. The reason that there is no congestion today is precisely because the cable companies have spent money in the past -- on DOCSIS 3, on logical node splits, on physical node splits. Plus, the fairness argument isn't exactly new.

SimbaSeven
I Void Warranties
join:2003-03-24
Billings, MT
·StarLink

1 edit

SimbaSeven

Member

said by tanzam75:

Yes, it's unfair that grandma doesn't get to pay $5 for cable Internet, while the bandwidth hog pays $1000. But surely it's fairer to charge grandma $50 and the bandwidth hog $100, rather than charging everyone $60.

That's why we have price tiers. If "grandma" wants to just check her email, stick her on the cheapest plan available (ours is 1.5mbps). If you're a power user, get the fastest one (ours is 30mbps). I don't see the blasted problem here.

Unfortunately, in some areas, that no longer applies. Now you have to buy a "byte" package and risk overages. This isn't "network management". This is just a new way of squeezing more money out of their customers.

..and if they can't handle multiple customers having super/ultra-fast tiers, why did they offer it in the first place?
tanzam75
join:2012-07-19

tanzam75

Member

said by SimbaSeven:

..and if they can't handle multiple customers having super/ultra-fast tiers, why did they offer it in the first place?

Because residential services are priced on the assumption that you will only burst to the max speed.

Max speed is not the only part of the equation. Average speed also matters. A 1.5 Mbps user who continually maxes out the connection is actually using more network resources than a 20 Mbps burst user who averages less than 1 Mbps.

Of course, the 10 Mbps user -- on average -- should use more data in a given month than the 20 Mbps user. So they ought to get a higher cap to go along with the higher speeds. But it still can't be unlimited, because the pricing of residential cable Internet is only made possible by the fact that people do not max out their connections.

SimbaSeven
I Void Warranties
join:2003-03-24
Billings, MT
·StarLink

SimbaSeven

Member

said by tanzam75:

Because residential services are priced on the assumption that you will only burst to the max speed.

Um.. If I'm paying for a 18mbps connection and I'm only getting 4, it becomes an issue. I should be getting at least between 12-18mbps constant, not whatever they feel like.
said by tanzam75:

Max speed is not the only part of the equation. Average speed also matters. A 1.5 Mbps user who continually maxes out the connection is actually using more network resources than a 20 Mbps burst user who averages less than 1 Mbps.

Depending on the technology. If you do it right, it shouldn't matter.
said by tanzam75:

But it still can't be unlimited, because the pricing of residential cable Internet is only made possible by the fact that people do not max out their connections.

This is the 21st century. More and more services are becoming internet-based instead. Also the bandwidth (mbps) is getting cheaper to deliver. The ISP's need to adapt or their customers will find other providers. Unfortunately, some people can't switch and are stuck with only a single ISP, that doesn't give a sh*t. There are a few that do, so I have to give them kudos.
tanzam75
join:2012-07-19

tanzam75

Member

said by SimbaSeven:

Um.. If I'm paying for a 18mbps connection and I'm only getting 4, it becomes an issue. I should be getting at least between 12-18mbps constant, not whatever they feel like.

But you're not paying for 18 Mbps of sustained bandwidth. You're paying for 18 Mbps burst, and "whatever they feel like" average speed.

Go to a Metro Ethernet company and see how much it costs to get 18 Mbps of dedicated bandwidth.

SimbaSeven
I Void Warranties
join:2003-03-24
Billings, MT
·StarLink

1 edit

SimbaSeven

Member

said by tanzam75:

But you're not paying for 18 Mbps of sustained bandwidth. You're paying for 18 Mbps burst, and "whatever they feel like" average speed.

Go to a Metro Ethernet company and see how much it costs to get 18 Mbps of dedicated bandwidth.

Did you read any of my post? I did not say I'm expecting 18Mbps all the time. I'm saying it should be at least in the same ballpark instead of half or less.

It's like me offering "up to" 100Mbps service (and charging for it), but only delivering 1/10 of that. Of course, to me, that's ethically wrong.
tanzam75
join:2012-07-19

tanzam75

Member

said by SimbaSeven:

Did you read any of my post? I did not say I'm expecting 18Mbps all the time. I'm saying it should be at least in the same ballpark instead of half or less.

Yes, I read all of it. You said that you "should be getting at least between 12-18mbps constant."

Since you said it should be "constant," then it should never fall below 12 mbps. That's dedicated bandwidth.

If you're saying that they should underpromise and overdeliver, then yes, that would be good business practice. Advertise a lower bandwidth than you're actually capable of providing, so that people get the advertised number much more often.

The problem is that you can never tell when a bandwidth hog will move in and wreck the calculations. Well, unless you put in caps, to keep the bandwidth hogs out.

And you're opposed to caps.

SimbaSeven
I Void Warranties
join:2003-03-24
Billings, MT
·StarLink

SimbaSeven

Member

said by tanzam75:

The problem is that you can never tell when a bandwidth hog will move in and wreck the calculations. Well, unless you put in caps, to keep the bandwidth hogs out.

And you're opposed to caps.

Caps don't do sh*t. It's just another way for them to make even more money without fixing the problem. Upgrading your infrastructure and backbone investment does.

..and the whole "bandwidth hog" thing is crap anyway. So, you have an OC3 connection feeding hundreds of customers that have purchased a 5mbps connection. Then a "bandwidth hog" comes in and uses a bit of it. Hey, he's paying for the 5mbps connection that you advertised. Then everyone's connection slows down because you failed to account for people actually using their connection speeds.

It's simple math. 5Mbps * 100 customers = 500Mbps. There is no way a single OC3 can handle that. You also have to account for growth if you get more customers. So if you get 150-200 customers, YOU NEED at least a GbE connection to handle that traffic, not the same OC3.

The point is, don't advertise the speeds you can't deliver. Sure, you can advertise "up to 1000Mbps", but if you can't deliver anywhere close to those speeds, your customers are going to hate you and there's the possibility of a lawsuit (false advertising).

Also, don't expect to minimally invest in your infrastructure and expect massive returns. I've seen plenty of ISP's and companies fail because of that.
tanzam75
join:2012-07-19

3 edits

tanzam75

Member

said by SimbaSeven:

..and the whole "bandwidth hog" thing is crap anyway. So, you have an OC3 connection feeding hundreds of customers that have purchased a 5mbps connection. Then a "bandwidth hog" comes in and uses a bit of it. Hey, he's paying for the 5mbps connection that you advertised. Then everyone's connection slows down because you failed to account for people actually using their connection speeds.

We're going around in circles here.
said by SimbaSeven:

It's simple math. 5Mbps * 100 customers = 500Mbps.

Again, you're talking about dedicated connections. A dedicated 5 Mbps connection would cost hundreds of dollars a month from a Metro Ethernet provider. (Although these days, they'd probably start you out at 10 Mbps.)

5 Mbps for 100 customers would never be provisioned at 500 Mbps in a residential setting -- more like 50. Which means that each user would have 500 Kbps of dedicated bandwidth. They can burst up to 5 Mbps, to download an ISO or watch a movie. What they cannot do is to use 5 Mbps all the time.

Average bandwidth consumption from residential customers is much, much lower than the burst demand. This is what makes oversubscription possible, and it is also why residential broadband can cost double-digits instead of triple-digits.

That 10x cost difference is supplied by the 10x oversubscription. That's what the caps are for. If you try to use a shared connection like a dedicated connection, then you should pay for it.