This is rather embarrassing. If OP had dedicated BW of 8MB I wouldn't be surprised if ISP started charging $200+/month for it. The whole reason why we share BW is because it's way cheaper and resources become more efficiently delegated as we have changes in network usage by different people.
Just go with the argument that $80/month is too expensive and explain why. Don't trifle with notion that we don't get dedicated BW.
Who said anything about dedicated bandwidth? There is nothing wrong with the ISP putting more than 1000 8Mbps users on an 8Gbps uplink. That's just basic network engineering.
But there is a difference between shared bandwidth and insufficient shared bandwidth. If your network's uplink is at in excess of 80% of capacity during peak usage periods, it's time to expand capacity. Follow that simple rule and all your 8Mbps customers will get at least 8Mbps, notwithstanding that they don't each have 8Mbps of dedicated bandwidth.
Obviously that doesn't include the rare outlier demand spikes, e.g. what happens to the network during a sudden disaster when many people all at once want to communicate with loved ones, but even there you won't have literally 100% of your customers using their full allocation at once. Which means that as long as you can predict within an order of magnitude the peak usage during such events, you can install capacity sufficient to carry it, which will be less than 100% dedicated bandwidth for everyone. And let's face it: We want the network to keep working during a disaster if at all possible; that's pretty important from a national security perspective (and therefore if nothing else from a public relations perspective for the carrier).
Granted installing capacity based on expected peak demand means that if people start using more bandwidth then more capacity will have to be installed and prices will have to increase. But so what? Isn't that how the market is supposed to function? Demand goes up, price goes up. If you don't need a specified amount of bandwidth then don't pay for it and subscribe to a slower package. But if you pay for it then you should be right to expect to get it.
2% ? that requires testing, information on TCP stacks used, computer used, distance, ping, network card, network drivers ...
And most importantly tested twice, and all other services on the computer have to be stopped. No messengers, no browsers, no updates zip nada nop.
Firstly, I'd bet dollars to donuts that his ISP advertises "up to" a certain speed and includes language in the T&Cs that specifies speeds may vary based on a variety of factors.
Secondly, is this even taking network overheads into account?
Thirdly, saying "kids in Africa might have 14kB/s for their whole connection, therefore that amount of bandwidth is worth fighting for" is dumb because it assumes constant marginal utility of bandwidth when that's obviously not the case. Going from 0kB/s to 14kB/s is a hell of a lot more valuable than going from 14kB/s to 28kB/s, and going from 14kB/s to 28kB/s is many orders of magnitude more valuable than going from 882kB/s to 896kB/s.
His ISP also likely only guarantees 70% of the advertised speeds be available and that the advertised speeds are theoretical maximums. This is what companies like AT&T and Charter do.
For what it's worth, most of the cost of providing a net connection is in the delivery to your endpoint ("last mile") and customer service. If all you want is X Mb/s and you can accept it at a place convenient for the ISP, prices are well under $10 per Mb/s per month. That is, a 100Mb/s bidirectional handoff will cost you significantly less than $1000.
Now, a 7Mb/750Kb connection? The return bandwidth can't be shared, so that gives you an idea of the contention ratio: about 1:10. The ISP is overselling their bandwidth ten times. This is probably reasonable for a consumer ISP. On the other hand, their cost for that bandwidth will be substantially under $7/month. The rest of the cost is going towards actually getting it to you, running a few services, customer interaction, and keeping everything repaired. Those other costs are relatively fixed -- it doesn't matter how many megabits you're buying, it costs the same per person -- but over-differentiating the prices seems to make sense to the marketing departments.
Pray for a sensible fiber provider, and in the meantime, don't let your local government abdicate its role in regulating telecom services.
What would ISP pricing look like if the market were heavily competitive (5+ alternatives) and the pricing was more like an electrical utility: $N just for the wired connection, and $B/gigabit after that?
I work at a place that provides dedicated bandwidth on fiber owned by the company. We would probably get results worse than he did. When you're trying to get to 100% theoretical utilization, you need to do alot of tuning, and then at best you'll pick up another 1%.
19 comments
[ 3.3 ms ] story [ 53.2 ms ] threadJust go with the argument that $80/month is too expensive and explain why. Don't trifle with notion that we don't get dedicated BW.
But there is a difference between shared bandwidth and insufficient shared bandwidth. If your network's uplink is at in excess of 80% of capacity during peak usage periods, it's time to expand capacity. Follow that simple rule and all your 8Mbps customers will get at least 8Mbps, notwithstanding that they don't each have 8Mbps of dedicated bandwidth.
Obviously that doesn't include the rare outlier demand spikes, e.g. what happens to the network during a sudden disaster when many people all at once want to communicate with loved ones, but even there you won't have literally 100% of your customers using their full allocation at once. Which means that as long as you can predict within an order of magnitude the peak usage during such events, you can install capacity sufficient to carry it, which will be less than 100% dedicated bandwidth for everyone. And let's face it: We want the network to keep working during a disaster if at all possible; that's pretty important from a national security perspective (and therefore if nothing else from a public relations perspective for the carrier).
Granted installing capacity based on expected peak demand means that if people start using more bandwidth then more capacity will have to be installed and prices will have to increase. But so what? Isn't that how the market is supposed to function? Demand goes up, price goes up. If you don't need a specified amount of bandwidth then don't pay for it and subscribe to a slower package. But if you pay for it then you should be right to expect to get it.
Firstly, I'd bet dollars to donuts that his ISP advertises "up to" a certain speed and includes language in the T&Cs that specifies speeds may vary based on a variety of factors.
Secondly, is this even taking network overheads into account?
Thirdly, saying "kids in Africa might have 14kB/s for their whole connection, therefore that amount of bandwidth is worth fighting for" is dumb because it assumes constant marginal utility of bandwidth when that's obviously not the case. Going from 0kB/s to 14kB/s is a hell of a lot more valuable than going from 14kB/s to 28kB/s, and going from 14kB/s to 28kB/s is many orders of magnitude more valuable than going from 882kB/s to 896kB/s.
Now, a 7Mb/750Kb connection? The return bandwidth can't be shared, so that gives you an idea of the contention ratio: about 1:10. The ISP is overselling their bandwidth ten times. This is probably reasonable for a consumer ISP. On the other hand, their cost for that bandwidth will be substantially under $7/month. The rest of the cost is going towards actually getting it to you, running a few services, customer interaction, and keeping everything repaired. Those other costs are relatively fixed -- it doesn't matter how many megabits you're buying, it costs the same per person -- but over-differentiating the prices seems to make sense to the marketing departments.
Pray for a sensible fiber provider, and in the meantime, don't let your local government abdicate its role in regulating telecom services.
98,4% in a speed test which highly relies on the destination server and the routing is a perfect result.
What would ISP pricing look like if the market were heavily competitive (5+ alternatives) and the pricing was more like an electrical utility: $N just for the wired connection, and $B/gigabit after that?
Per your methodology we have adjusted your rate based upon services rendered. We are also adjusting your bill based upon exemplary upload performance.
((6.89/7)+(0.78/.75))/2 * 80
Amount Due: $80.97
-Your customer
I work at a place that provides dedicated bandwidth on fiber owned by the company. We would probably get results worse than he did. When you're trying to get to 100% theoretical utilization, you need to do alot of tuning, and then at best you'll pick up another 1%.