The Busy Hour and Data Caps

As states are getting ready to create their broadband plans for the NTIA’s $42.5 billion BEAD grants, we’re starting to see some interesting arguments being made by incumbents to influence state broadband plans. One of the aspects of the BEAD plan that hasn’t been discussed much yet is that the NTIA is stressing affordability. For example, the NOFO states several times that states must develop a middle-class rate plan. Everybody I know is scratching their head on that what that means, but to the big ISPs, this must sound like rate caps – something they have vigorously fought everywhere.

One of my readers from Maine says that one aspect of high rates – data caps – is a big topic of discussion in the state. ISPs are making the claim that they’ve made many times that data caps are needed to manage the network. The rest of the ISP argument is that heavy broadband users are creating extra costs and should pay for the extra usage.

I’ve written about this before, and the big ISP argument is pure bosh. Broadband costs are not related to the overall volume of broadband being delivered on a network. The cost is determined almost entirely by what network engineers call the busy hour. The business hour each day is when a network sees the busiest broadband usage. For networks with a lot of residential customers, the busy hour is usually during the evening when the most families are streaming video. However, I’ve seen a few networks where the busy hour was at some different time of the day. One example I know of is a network serving college students that has more usage after 10:00 PM and into the early hours of the morning, which the ISP accredits to gaming. Networks that serve business customers might have the busy hour at almost any time of the day, depending upon the mix of businesses using the network.

The busy hour is important because it determines how much bandwidth is needed for the various components of the network. For example, the data connection between the ISP core and a neighborhood must be large enough to handle the busy hour – otherwise, bandwidth and speeds get slower. Another important network segment that depends upon the busy hour is the connection to and from the ISP and the Internet – there must be enough bandwidth available to accommodate the expected busiest time each month. Most ISPs buy enough bandwidth to the Internet to have at least a 20% cushion of capacity greater than the expected busiest hour.

As long as an ISP is buying enough broadband to the Internet to satisfy the busy hour, it doesn’t matter how busy the network is the rest of the time – because all other times are less busy than the peak time. The ISP is paying for the peak bandwidth and doesn’t get any savings in bandwidth cost for times when customers don’t use the bandwidth.

The data cap for most ISPs is set at or close to 1 terabyte per month (1,000 gigabytes) of total combined upload and download usage for a customer. According to the most recent statistics from OpenVault, 15% of all homes are now using more than a terabyte of bandwidth per month. I see network performance statistics from my clients, and none of them believe that the busy hour is caused by a few big users – it happens at the time of the day when the highest number of homes use broadband at the same time. There is a limit on how much bandwidth a given home can realistically use if they are doing the same things that many homes do – like watching video or gaming. A house that uses a terabyte of data per month is likely no busier than a home that uses half that. The house with the higher monthly usage normally just uses the broadband for more hours per day to accumulate the terabyte of usage.

The only time that data usage costs more money for an ISP is when it has to increase the size of the data connection to the Internet. But interestingly, even that doesn’t always cost an ISP more money because as network usage has grown, the cost per gigabit of traffic to reach the Internet has dropped year after year. I have ISPs that are carrying four times the traffic of only a few years ago but are still paying no more for the connection – which is another argument about heavy users creating more costs for an ISP.

Data caps do not make any significant change to lower the busy hour, and any ISP that says they do is being untruthful. Data caps serve only one purpose – to make additional money for ISPs. There is no other explanation, and any ISP that argues that data caps are a network control issue is counting on the technology naivety of regulators and policy people. When Comcast first introduced data caps, only a fraction of one percent of homes used more than a terabyte per month. Comcast made the same argument then that the data caps were to rein in the usage of the few heavy users. The argument was no more valid then than it is now. ISPs have to be thrilled to see so many homes now exceeding the data caps – the overage fees for exceeding the data caps can be as much as $50 per household. Seeing 15% of households exceeding the data caps is like Christmas to big ISPs, like a big secret rate increase.

One thought on “The Busy Hour and Data Caps

  1. “The only time that data usage costs more money for an ISP is when it has to increase the size of the data connection to the Internet.”

    So if edge storage and processing were used to time-shift some traffic from the busy hour to other times, the result would be a decrease in the overall cost of connectivity? For instance if videos that were preordered and downloaded to edge storage were not charged against transmission caps? Or if software updates were buffered in edge storage and downloaded across the backbone when traffic is lower?

Leave a Reply