3 ms·

>Metered Internet access is potentially reasonable Except when its not, because everyone will bitch & moan about rocking the boat. When Comcast added the 250GB

>Metered Internet access is potentially reasonable Except when its not, because everyone will bitch & moan about rocking the boat. When Comcast added the 250GB monthly limit, there was a large vocal minority that made sure that that restriction never left trail.

Dylan16807 13y ago
That's because 250GB is stupidly low. It equates to only 4% utilization on a 20mbps connection. I can buy 250GB in a datacenter for a dollar. And on top of that the cap didn't grow over time to reflect increased bandwidth needs and decreasing bandwidth costs. Give me a cap that represents fair bandwidth prices and I won't object.
- kbenson 13y ago
  250GB may be low, but if you think that you can equate data center prices to prices for transit delivered to an end-user's address, you aren't thinking hard enough.
  - Dylan16807 13y ago
    Educate me. As far as I know, the only expensive part is in establishing a connection to a user. The cost of extra data is only in the backbone parts of the network, and those do not eat a large portion of revenue.
    - res0nat0r 13y ago
      Old, but very detailed: http://drpeering.net/white-papers/A-Business-Case-For-Peering.php http://drpeering.net/white-papers/A-Business-Case-For-Peerin...
    - kbenson 13y ago
      In short, data centers are generally located at the core of networks, end users are, by definition, at the fringe of the network. All the equipment and lines between the core and the end user add to the cost to supply bandwidth through up-front purchasing cost, maintenance cost and configuration cost. All of those are costs that would not have to be paid if they did not supply internet to end users. To assume that the cost of just supplying bandwidth to a date center is the same as through all that infrastructure is to assume the cost of all that is zero, or close enough they write it off, while in truth it probably accounts for the lion's share of their overhead.
      - nknighthb 13y ago
        aka the last mile problem. http://en.wikipedia.org/wiki/Last_mile http://en.wikipedia.org/wiki/Last_mile
        Dylan16807 13y ago
        ...except that the last mile has the same 10mbps or 50mbps or whatever capacity even if you're on the crippled-speed budget plan. The only cost to increase the quota is for the ISP to upgrade its major inter-site links, not the edge connections. The last mile limits bitrate, not quota.
        kbenson 13y ago
        If even 10% of the end users fully utilized their circuits they would be far over their capacity to get those packets to their core, even if the core could handle moving them to other networks. As such, increasing the amount of transit the ISP purchases may do nothing for allowing it to be effectively utilized at the end user location.
JoshTriplett 13y ago
First of all, that's ludicrously low. More importantly, it's a new restriction on an existing service, rather than a new service. Both of those lead to push-back from current and potential customers. In any case, the problem is that Google has already pitched their fiber service as gigabit, and attempting to meter it will make it look less appealing. Personally, I find arbitrary restrictions on the type of data I can transfer even less appealing, but many people won't. What would Google say to someone saturating their bandwidth 24 hours a day and not running a server? I can easily think of ways to do that. For instance, how about doing daily image backups of your home media center?

Dylan16807 13y ago

That's because 250GB is stupidly low. It equates to only 4% utilization on a 20mbps connection. I can buy 250GB in a datacenter for a dollar. And on top of that the cap didn't grow over time to reflect increased bandwidth needs and decreasing bandwidth costs. Give me a cap that represents fair bandwidth prices and I won't object.

kbenson 13y ago
250GB may be low, but if you think that you can equate data center prices to prices for transit delivered to an end-user's address, you aren't thinking hard enough.
- Dylan16807 13y ago
  Educate me. As far as I know, the only expensive part is in establishing a connection to a user. The cost of extra data is only in the backbone parts of the network, and those do not eat a large portion of revenue.
  - res0nat0r 13y ago
    Old, but very detailed: http://drpeering.net/white-papers/A-Business-Case-For-Peering.php http://drpeering.net/white-papers/A-Business-Case-For-Peerin...
  - kbenson 13y ago
    In short, data centers are generally located at the core of networks, end users are, by definition, at the fringe of the network. All the equipment and lines between the core and the end user add to the cost to supply bandwidth through up-front purchasing cost, maintenance cost and configuration cost. All of those are costs that would not have to be paid if they did not supply internet to end users. To assume that the cost of just supplying bandwidth to a date center is the same as through all that infrastructure is to assume the cost of all that is zero, or close enough they write it off, while in truth it probably accounts for the lion's share of their overhead.
    - nknighthb 13y ago
      aka the last mile problem. http://en.wikipedia.org/wiki/Last_mile http://en.wikipedia.org/wiki/Last_mile
      - Dylan16807 13y ago
        ...except that the last mile has the same 10mbps or 50mbps or whatever capacity even if you're on the crippled-speed budget plan. The only cost to increase the quota is for the ISP to upgrade its major inter-site links, not the edge connections. The last mile limits bitrate, not quota.
        kbenson 13y ago
        If even 10% of the end users fully utilized their circuits they would be far over their capacity to get those packets to their core, even if the core could handle moving them to other networks. As such, increasing the amount of transit the ISP purchases may do nothing for allowing it to be effectively utilized at the end user location.

JoshTriplett 13y ago

First of all, that's ludicrously low. More importantly, it's a new restriction on an existing service, rather than a new service. Both of those lead to push-back from current and potential customers. In any case, the problem is that Google has already pitched their fiber service as gigabit, and attempting to meter it will make it look less appealing. Personally, I find arbitrary restrictions on the type of data I can transfer even less appealing, but many people won't. What would Google say to someone saturating their bandwidth 24 hours a day and not running a server? I can easily think of ways to do that. For instance, how about doing daily image backups of your home media center?