We know where the 20 Mbps number comes from. Early discussions of the BEAD grant rules considered that grant-eligible areas would be anywhere that customers don’t have the option to purchase broadband with speeds of at least 100/100 Mbps. But cable companies and the wireless industry went ballistic, and there was furious lobbying to lower the BEAD coverage definition to 100/20 Mbps. The reason for this was obvious since both technologies at the time could meet the 100 Mbps download speed but could not meet a 100 Mbps upload speed requirement.
This means that the 20 Mbps definition is a political compromise that has nothing to do with the broadband speeds that households and businesses need. Interestingly, technology in those two industries is changing quickly, and cable companies and WISPs are on the verge of being able to meet 100 Mbps uploads if they upgrade to the latest technology.
Several of the big cable companies are currently implementing mid-split technology upgrades that I’ve seen reported as delivering from 100 Mbps to 300 Mbps upload speeds depending upon the local conditions in a cable company network. The big cable companies have all said that they intend to implement DOCSIS 4.0 upgrades that will enable gigabit upload speeds. But cable companies will likely continue to fight to keep the 20 Mbps definition because they will not want to upgrade their networks in smaller and non-competitive markets.
We’re also on the verge of big changes in fixed wireless technology. As WISPs implement the newest radios, and particularly when they integrate 6 GHz spectrum, the networks will be able to deliver much faster speeds. Wireless network technology is interesting in that the ISP can determine the amount of upload and download speed to offer – and as overall speeds get faster, WISPs will be able to deliver 100 Mbps speeds if they elect to do so.
We’re also seeing speed increases from FWA cellular wireless. Verizon recently reported the ability to deliver much faster speeds with the introduction of C-Band spectrum into towers. If they choose, Verizon and the other carriers using C-Band could also meet a 100 Mbps upload speed.
The FCC should consider a faster definition of broadband just because of the changes in technology. There is no reason to set a low definition of 20 Mbps upload that only rewards ISPs that want to stick with older technology. To do so is reminiscent of past FCC decisions that protected DSL long after it was obsolete.
What I find puzzling is that in the NOI, the FCC argued for a faster definition of upload speeds. They mention a study by the Consortium School Network (CoSN) that says that a single student working at home should have a 12 Mbps connection and a 20 Mbps connection is not sufficient to allow multiple students to work from home. The FCC also acknowledges that there are a lot of uses of broadband today that need faster connections. For example, the NOI cites that a 25 Mbps connection is needed for 4K video conferencing, telehealth, and remote learning. The FCC cites that graphics-intensive work can require a 45 Mbps connection. And they acknowledge what millions of gamers will tell you, which is that a 20 Mbps connection is far from adequate. Keeping a 20 Mbps definition of broadband is a regulatory decision that says that these faster uses of broadband are not important or needed.
We also can’t forget that the definition of broadband is not just for households. Cable company networks that offer 20 Mbps upload to businesses are massively inadequate. Businesses have migrated a lot of functions to the cloud, and I could list fifty ways that businesses want to use upload broadband – and many can’t due to slow technologies.
You might think that the definition of broadband is not important – but it says that any ISP connection slower than the definition is not really broadband. If the FCC considers a faster upload speed it will provide an incentive for ISPs to upgrade technology to meet a faster definition – since customers will demand speeds that are considered to be broadband.
Unfortunately, the FCC will likely adopt the 20 Mbps definition, and millions of homes, and particularly businesses, will continue to suffer with inadequate upload speeds for the next five years until the FCC looks again at the definition of broadband. This is a chance for the FCC to implement a policy change that will have real market implications. But the FCC probably doesn’t want to face the ISP lobbying effort to keep an inadequate definition of broadband – even as ISPs are already making upgrades that can meet a faster definition.
6 replies on “The Definition of Upload Speed”
As I am perpetually pointing out, improving network queue management along the way, would have a far, far greater effect on improving the network performance than changing the speed, while lowering costs to the end users. There is so much evidence accumulated of this that I still boggle at demands for more speed, instead of merely setting the buffering right for the bandwidth available. What would I take to get you to run a simple test, such as the waveform test, that would demonstrate to you that perhaps, perhaps, myself and other founders of the internet have been making a point about reducing latency, rather than worrying about bandwidth?
Try this: https://blog.cloudflare.com/making-home-internet-faster/
I also have to question the circular argument that each student needs 12mbit uploads, also. No combination of traffic I can think up, even the upload of a paper, can justify a need for more than 10Mbits, total, per household, and not per student, in this case.
There used to be an argument in favor of symmetry in that many of our baseline technologies can do that in hardware, not needing a CPU intensive shaper to do it. However asymmetry is built into GPON and modern fiber and wireless also, in modern hardware, and hard to change. IF only vast improvements in upload speed were accompanied accompanied by a shift back to using more local compute, I would love it!
The NOI comment period ends December 1, and you make many compelling arguments for your side, that while I do not agree with many, suggest you submit more formally to the FCC process. I very much question the reliance on the dataset they demand to be used, and wish very
I have to re-inforce much of what Dave said here. We have a lot of customers on ’25×5′ provisioned at 30×7. We’ve implemented QoE from two companies, preseem and libreqos, on various parts of our network.
Once we implemented this, we have a lot of struggle getting people to upgrade from these plans. 30M download is enough to stream multiple 4K videos and use the internet which is easily enough for most people. 7M up is easily enough to stream a zoom call, send an email, and do typical iphone and pc backups. I say ‘easily’ but that’s with QoE that keeps buffer bloat from causing self-inflicted congestion issues.
I do think the ‘standard’ number needs to be a bit higher, new media I think pushes this up to around 40M down and 15M up or so, I would be happy with a 50×20 really.
The push for more and more speed in the absense of ‘QoE’ or a wholistic view on network performance slows adoption of ‘broadband’ for so many people. They are stuck waiting with 3-5Mbps DSL because operators can’t build moderate speeds to them, they have to focus their money on high speeds that many just don’t NEED. They aboslutely NEED modest speeds with good quality and the current policies are delaying universal access to modest services because they focus on premium products.
Cables can deliver a much higher upload speed to most user quickly ，100Mbps above， but they do not want to do that！ Bad guys！
It’s really not that easy. Lots of coax plants were built back in DOCSIS 1.1/2.0 days, then upgraded to 3.0 and then PARTIALLY upgraded to 3.1. That 3.1 upgrade was primarily fiber to the node and a new remote CMTS but using the old RG59 cable to the homes which is often not able to modulate anywhere near max capacity. You end up with a 4:1 split on channels most of the time so they can sell the biggest download possible (their own marketing trap) and 35Mbps really is about the top end of what is deliverable. They also don’t have as high a modulation in the upstream so it’s not 4:1 throughput ratio, that’s channel mix, and usually means a least a 10:1 throughput ratio. They could probably offer a little bit more if the cable tests out well enough but they are big companies and having simple options is pretty much a necessity. If most of your customers cant get much over 35, that’s what you offer.
The Cable industry very much needs DOCSIS 4.0 which is released but sort of hard to get your hands on. It never caught on in a big way. It does have equal modulations in the upstream though and better noise handling, it would be a ‘drop in’ replacement and they’d get more out of the RG59 runs. They’d still need to upgrade the drops to RG6 for homes that wanted top speeds, ie to compete with these XGSPON operations offering <=8Gbps. These companies also have hundreds of thousands of DOCSIS 3.1 modems out there to replace, it's a huge undertaking that might take too long for them to compete with the fttx startups.
Probably the lack of real competition has kept cable cos from upgrading more rapidly, but all this funding for fiber is really putting the pressure on. I've heard that Spectrum has lost about 10,000 subs in my town of 150,000 to TDS as they've built in and they're only servicing about 1/3 of town.
I can't imagine the cable cos would just happily shed those numbers if they could magically turn up the speeds.
The real reason behind is that cables want to
constrain and delay the demand for the application based on the higher upload speed. After the rise of new demand, users will need a higher quantity of network, the cables network will lose the competition with fiber network more quickly!
Sorry!
need the network with a higher quanlity .