Cox → [NV] Received email of a Cox Data Usage Notification

The last sentence you put there is the key issue. Bandwidth is cheap, if you have access to a tier 1 provider. At the last mile though the price goes way up.

Why is that? Well, try laying single mode fiber all the way from downtown phoenix to your house. Even if you could afford all of the equipment and raw materials, the government would also want to throw a few regulations up your ass and a good measure of taxes here and there.

The price lock guarantee is the residential contract, which includes an ETF.

If you don't enforce it, heavy users might get too comfortable just filling up their pipe willy nilly. The idea behind the cap is to keep it so that they'll be conscious about downloading that 42GB blu-ray rip of the dark knight. If they do this too much, they can impact the performance of other users. Ideally your ISP wants to provide their customers with full bandwidth 24/7.

In my opinion, there is a better solution to the cap, but it would probably require an investment in additional hardware (possibly even additional CPE hardware, e.g. a new cable modem) and maybe some statisticians to develop a proper algorithm. If you get a heavy user who frequently uses up a lot of data, then you throttle them *only* during peak hours, and at a statistically appropriate amount based on how congested the node is (not some arbitrary amount, e.g. none of the "we'll drop you by half" BS that some providers do) and QoS them at all other times, also separating bulk transfers from burst transfers. This makes it so that they have no cap, but they're going to see reduced download speeds at certain times of the day and week, and they are last in line to have their packets delivered. Ideally this should be done with no noticeable increase in packet delay (aka latency, ping times.)

I'd say something like a policy whereby if they hit their monthly cap, then for that month and the next month they will be put into this category of customer, and notified of it. If they hit the cap again in the next month, they stay on it, and be notified of it.

This would allow them to be cap free, without harming other customers.

I'd be fine with being subject to such a policy myself, provided there is no observable increase in packet delay, and the throttling does a statistically intelligent and reasonable decrease rather than an arbitrary one.

I'm about 8 weeks in--past the 30 days.

I'll try again, maybe I'll get a different answer with a different rep. I'll mention Price Lock specifically.

I know of no way to to do your idea without increasing ping times. You are asking someone to interfere with the travel of something, without affecting its travel time. You cannot have your cake and eat it at the same time. If you want to get the attention of an excessive user, a big significant reduction in download and upload speeds, with huge ping time increase, will do it in many cases. Since many of the worst offenders are gamers, a huge increase in latency, each time they are playing, will help remind them to be more responsible in the future. If you leave them with the regular latency, they will not care much about reduced upload and download speeds, unless it is a very large decrease.

Your idea of rapid analysis for QoS is already available in the network gear supplied by Sandvine and others. The ISP has to spend enough on the network equipment to give enough headroom to do the analysis more frequently. Uh oh! ISPs today are not spending enough on network upgrades to provide decent subscriber experiences. Why would they spend an a large amount to rapidly, and with great finesse, provide QoS? It is easier and less costly to reduce speeds in big percentages for set time periods to reduce congestion or punish excessive users.

Years ago, researchers in Texas established through scientific analysis that all these fancy usage control schemes were inferior to upgrading the network continuously. They showed that, over a time period of a few decades, you could spend the same amount to do two different schemes. One was monitoring, restricting, punishing, and limiting users. The other was to continuously improve the network using upgrades in fiber optic, coaxial cable, and Ethernet networking technologies. One provided limited increases in network usability, but huge profits to sellers of punishment technology. The other provided statistically significant yearly increases in network usability and profits to those advancing network technology. Guess which scheme most ISPs decided to use? Guess which scheme appeals most to Wall Street?

Of course you'll effect it, the question is by how much, hence why I say noticeable. An IPv6 solution would be to tell the MTU Path Discovery that it has a smaller MTU than it really does (fragmentation isn't allowed in IPv6 - the stack has to probe the path and figure out the optimal MTU to the destination before opening a socket.) Smaller packets can be throttled with less delay. There is the issue of decreasing the MTU of a socket already opened, however those can just be handled with normal delay, and all new ones get a smaller MTU. If a gamer runs into problems with that, simply reconnecting to the server will fix it.

Combined with QoS and tiering, you could give real-time applications low latency while throttling. I already have low latency without throttling my usenet downloads (which completely saturate my downstream bandwidth) thanks to the amazing QoS capabilities of tomato firmware. Companies like Cox already have DPI in place from the days when they used to insert the reset flag into TCP packets of edonkey filesharers, so I'm sure it could be put to work here.What kind of networks did they look at? Not all technologies are created equal, and I am dubious that they could have examined all scenarios.

Anyways you work under the assumption that there's some guy sitting behind a big desk plotting his next dastardly deed to hurt broadband deployments. Maybe there's a real Professor Chaos or Dr. Evil out there who are evil just for the sake of being evil, but chances are none of them are running a broadband provider. If the later was still profitable, I'm sure they'd go for it to outpace the competition. Any idiot will tell you that stagnation leads to bankruptcy.

The fact is, no matter how much you increase the available bandwidth in the network, you'll still end up with people who treat their pipe as if they are running a cloud service on a last mile connection. You will have to deal with them one way or another. I'm sure even your scientists in Texas would agree that last mile connections aren't ideal for cloud services, and that these guys will have to be dealt with one way or another.

What really sucks about this whole thing, is that your upload is being calculated into the 250GB limit for Premiere. I downloaded 1.5GB yesterday, checked the data meter this morning and it said i downloaded 19.56GB, which is bunk . because i uploaded 18GB but only downloaded 1.5GB. So my question is if the meter says download, why is it adding the upload and download together and taking away from the monthly 250GB limit, but only calling it download???

Much of the terminology has been simplified for the benefit of less technical users. To answer the question the monthly allowances are for combined upload and download.

And the crazy part is, its all metered the same. There's an abundance of 8 QAM256 channels on the node, not so much on the upstream side of things.

Thanks for letting me know that. I tried again, this time with a support rep instead of a sales rep. The support rep could not help, but did put me through to a "retention" rep.

The retention rep upgraded me to Premiere, under the same discount percentage as my previous commitment.