Data caps do not help congestion in any form and discourages new legal usages. It has no beneficiary use outside of a few exceptions like a satellite link or other limited connection. Throttling typically costs more than adding new bandwidth, except when a networks is horrible miss-managed or the ISP cannot afford to purchase large enough bulk bandwidth to get decent rates.
vLANs are a Level 2 layer that has Ethernet ride on top of it, no issues as it is clean and maintains standards. What I'm not a fan of is not following standards. The first and last IP address of a subnet are reserved, this means a/31 and/32 cannot follow the standard.
Of the people capable of tinkering with graphic drivers, they're all hired by AMD, nvidia, or Intel. Everyone else has no idea what they're doing in there.
Your work arounds don't follow the IPv4 standard. Using standards complaint IPv4 Ethernet hardware, how does one not use at least a/30?/31 and/32 are not valid. They're like NAT. A horrible bandaid.
QoS is more expensive to implement than adding bandwidth in most cases. If you're going to offer artificially limited services, data-caps are the way to go.
The amount of data you use has no bearing on cost or load, only when you use or how fast you burst. Quick high-speed bursts are much easier and cheaper than long drawn-out transfers. You essentially have to use 95th percentile to effectively charge people, but you can't pre-pay on 95th percentile, because you won't know what it is until after the billing period. Then you have the whole issue of monitoring every customer and trying to explain to them how it works
Even then, someone consuming a steady 1Gb/s from midnight to 11am will still cost an ISP less than someone transferring 1Mb/s for 2 hours during 5pm-11pm.
Essentially, it comes down to that the individual does not matter, only the group as a whole matters. So if the group determines costs and not the individual, you can't charge people individually, but as a group.
What ISPs are doing to high data usage customer is that if you are too many sigmas from the norm, then you're not longer part of the group, so you cannot get the group rate. Instead you get the individual rate, which is much more expensive, aka Business customers.
You can over-sell your back-haul as data usage at large trunks is very stable, even when highly over-subscribed. It's the last-mile that you can't over-subscribe because there are not enough users at those points to make a smooth average.
The problem is a fair system would be based on the 95th percentile, and the average user would have no idea how to measure it. But I would love to have an option of an opt-in 95th percentile where I can do my own traffic shaping, then they can sell me bandwidth with a minimum commit. If I make full use of my 1Gb, charge me the full $1,000 of bandwidth. Won't matter much in the long run, because bandwidth prices drop 50% every year, so $1,000 this year will be only $500 next year.
I may rather have a 100Mb connection and be charged by 95th percentile and run a server without and SLA, than a 1Gb connection where I can't run a server.
many ISPs have started blocking inbound port 25 across the board for no particular reason
An opened port 25 is the #1 cause of spam. Now that it has been blocked by most ISPs, it's a non-issue, but opening it back up for everyone would dramatically increase the spam to the point of it being the #1 cause again.
And how do you plan on routing data to the end user if the end user doesn't have their own subnet? Do you plan on using one large subnet and putting hundreds of customers in the same broadcast domain so they can easily probe each-other out? That would be a security nightmare.
There is a noticeable difference between servers and regular usages. I look at my work's network usage and the output can hold a steady 850Mb/s with only minor ripples, humps, and spikes. If I un-throttle my symmetric connection at home, even with the fastest of people downloading from me, it doesn't take long and there are huge troughs and crests.
The difference between a real server a play-server is that a real server is very popular and has thousands of clients constantly transferring data, while a lesser play-server will not be so popular and tends to be all over the place.
Net Neutrality makes an exception for using your connection in a way that degrades other user's experience, with in reason. That is up to your ISP to decide and handle. If you want to pay more money for a business class connection that is meant to handle these concerns, then the ISP can allow it.
In other words, if your usage is too many sigmas different than the norm, then you're the nail that's standing out and may get hammered into place.
or non-harmful devices [subject to reasonable network management]
I bet servers fall under "harmful devices". I think the issue is that allowing servers opens the flood-doors for many business to drop their dedicated lines and go residential. Most businesses that already have a dedicated line will probably stick with it because they need the SLA and other services, but there are still a lot of businesses they don't need an SLA but just gobs of bandwidth.
Google may be better off saying that servers are allowed, but if you are using lots of bandwidth, you may need to do some traffic shaping in order to make your server be non-harmful. Then Google can address the issues as the arise.
"Harmful" servers should be easy to detect and should be a corner case.
The uninsured are basically screwed, and are asked to pay many times what is charged to the insured or medicaid patients.
There are ways around not having insurance. I was talking to an ER nurse who had been working in the ER for over 20 years. He said he sees people who come into the ER once per year, claiming to have a slew of issues. The ER is required by law to check them out, insurance or not.
The ER is also several times more expensive than someone scheduling a regular appointment. The nurse said so much money could be saved if these people could get a regular appointment than going through the ER, but because of no insurance, it's the only way.
Since these people don't have much money, the hospital eats the costs because they know they can't squeeze blood from a rock, and they raise the rates for everyone else who can pay.
Luckily our hospital system is one of the best in the nation/world, so people come from out of state or country to stop in for special treatments. The hospitals charge out-of-state people more than in-state, by quite a bit. This helps subsidize many of these in-state losses.
I can see military applications. With all the tools to already calculate air resistances and trajectories, what's to stop someone from picking a target anywhere on Earth?
I can be sympathetic to their up-bringing that contributed to their bad choices, but it doesn't mean I'll feel entirely sorry for a murderer that had a bad up-bringing.
One of those "maybe in another life time, we could have been friends", but not in this one.
What we see as illness in an individual and seems a bad survival trade, is sometimes a good survival trait for the entire DNA family. You can't look at a person as ask how it benefits their DNA's propagation, but how their DNA may help the DNA of their blood relatives propagate.
Slashdot Mirror
Data caps do not help congestion in any form and discourages new legal usages. It has no beneficiary use outside of a few exceptions like a satellite link or other limited connection. Throttling typically costs more than adding new bandwidth, except when a networks is horrible miss-managed or the ISP cannot afford to purchase large enough bulk bandwidth to get decent rates.
vLANs are a Level 2 layer that has Ethernet ride on top of it, no issues as it is clean and maintains standards. What I'm not a fan of is not following standards. The first and last IP address of a subnet are reserved, this means a /31 and /32 cannot follow the standard.
Of the people capable of tinkering with graphic drivers, they're all hired by AMD, nvidia, or Intel. Everyone else has no idea what they're doing in there.
Your work arounds don't follow the IPv4 standard. Using standards complaint IPv4 Ethernet hardware, how does one not use at least a /30? /31 and /32 are not valid. They're like NAT. A horrible bandaid.
And what might those be? Don't say data caps or throttling.
QoS is more expensive to implement than adding bandwidth in most cases. If you're going to offer artificially limited services, data-caps are the way to go.
What about Pentium 4s?
The amount of data you use has no bearing on cost or load, only when you use or how fast you burst. Quick high-speed bursts are much easier and cheaper than long drawn-out transfers. You essentially have to use 95th percentile to effectively charge people, but you can't pre-pay on 95th percentile, because you won't know what it is until after the billing period. Then you have the whole issue of monitoring every customer and trying to explain to them how it works
Even then, someone consuming a steady 1Gb/s from midnight to 11am will still cost an ISP less than someone transferring 1Mb/s for 2 hours during 5pm-11pm.
Essentially, it comes down to that the individual does not matter, only the group as a whole matters. So if the group determines costs and not the individual, you can't charge people individually, but as a group.
What ISPs are doing to high data usage customer is that if you are too many sigmas from the norm, then you're not longer part of the group, so you cannot get the group rate. Instead you get the individual rate, which is much more expensive, aka Business customers.
You can over-sell your back-haul as data usage at large trunks is very stable, even when highly over-subscribed. It's the last-mile that you can't over-subscribe because there are not enough users at those points to make a smooth average.
it should be about your traffic profile
The problem is a fair system would be based on the 95th percentile, and the average user would have no idea how to measure it. But I would love to have an option of an opt-in 95th percentile where I can do my own traffic shaping, then they can sell me bandwidth with a minimum commit. If I make full use of my 1Gb, charge me the full $1,000 of bandwidth. Won't matter much in the long run, because bandwidth prices drop 50% every year, so $1,000 this year will be only $500 next year.
I may rather have a 100Mb connection and be charged by 95th percentile and run a server without and SLA, than a 1Gb connection where I can't run a server.
many ISPs have started blocking inbound port 25 across the board for no particular reason
An opened port 25 is the #1 cause of spam. Now that it has been blocked by most ISPs, it's a non-issue, but opening it back up for everyone would dramatically increase the spam to the point of it being the #1 cause again.
is wasteful, subnets even more so.
And how do you plan on routing data to the end user if the end user doesn't have their own subnet? Do you plan on using one large subnet and putting hundreds of customers in the same broadcast domain so they can easily probe each-other out? That would be a security nightmare.
There is a noticeable difference between servers and regular usages. I look at my work's network usage and the output can hold a steady 850Mb/s with only minor ripples, humps, and spikes. If I un-throttle my symmetric connection at home, even with the fastest of people downloading from me, it doesn't take long and there are huge troughs and crests.
The difference between a real server a play-server is that a real server is very popular and has thousands of clients constantly transferring data, while a lesser play-server will not be so popular and tends to be all over the place.
The network patterns are completely different.
A proper ToS that included network management would require both a lawyer and a network admin to understand it.
Net Neutrality makes an exception for using your connection in a way that degrades other user's experience, with in reason. That is up to your ISP to decide and handle. If you want to pay more money for a business class connection that is meant to handle these concerns, then the ISP can allow it.
In other words, if your usage is too many sigmas different than the norm, then you're the nail that's standing out and may get hammered into place.
They don't care what you use it for, just the network load patterns commercial servers create.
or non-harmful devices [subject to reasonable network management]
I bet servers fall under "harmful devices". I think the issue is that allowing servers opens the flood-doors for many business to drop their dedicated lines and go residential. Most businesses that already have a dedicated line will probably stick with it because they need the SLA and other services, but there are still a lot of businesses they don't need an SLA but just gobs of bandwidth.
Google may be better off saying that servers are allowed, but if you are using lots of bandwidth, you may need to do some traffic shaping in order to make your server be non-harmful. Then Google can address the issues as the arise.
"Harmful" servers should be easy to detect and should be a corner case.
The uninsured are basically screwed, and are asked to pay many times what is charged to the insured or medicaid patients.
There are ways around not having insurance. I was talking to an ER nurse who had been working in the ER for over 20 years. He said he sees people who come into the ER once per year, claiming to have a slew of issues. The ER is required by law to check them out, insurance or not.
The ER is also several times more expensive than someone scheduling a regular appointment. The nurse said so much money could be saved if these people could get a regular appointment than going through the ER, but because of no insurance, it's the only way.
Since these people don't have much money, the hospital eats the costs because they know they can't squeeze blood from a rock, and they raise the rates for everyone else who can pay.
Luckily our hospital system is one of the best in the nation/world, so people come from out of state or country to stop in for special treatments. The hospitals charge out-of-state people more than in-state, by quite a bit. This helps subsidize many of these in-state losses.
I can see military applications. With all the tools to already calculate air resistances and trajectories, what's to stop someone from picking a target anywhere on Earth?
Even the P4 had clock cycles short enough that they had to add stages in the pipeline to allow the signal to propagate across the chip.
Everything stopped being funny YEARS ago
Welcome to being old.
Since space and time didn't exist prior to the Big Bang, rotation would have no meaning because rotation is a rate and rates require time.
Their new MMORPG that they have been working on for years and doing massive hiring for, is entirely new IP.
I can be sympathetic to their up-bringing that contributed to their bad choices, but it doesn't mean I'll feel entirely sorry for a murderer that had a bad up-bringing.
One of those "maybe in another life time, we could have been friends", but not in this one.
What we see as illness in an individual and seems a bad survival trade, is sometimes a good survival trait for the entire DNA family. You can't look at a person as ask how it benefits their DNA's propagation, but how their DNA may help the DNA of their blood relatives propagate.