The marketing are of the broadband industry spends a lot of time convincing folks that the most important part of a broadband product is download speed. This makes sense if fiber of cable are competing in a market against slower technologies. But it seems like most advertising about speed is to convince existing customers to upgrade to faster speeds. While download speed is performance, the industry doesn’t spend much time talking about the other important attributes of broadband.
Upload Speed. Households that make multiple simultaneous upload connections like video calls, gaming, or connecting to a work or school server quickly come to understand the importance of upload speeds if they don’t have enough. This was the primary problem that millions of households subscribed to cable companies encountered during the pandemic when they suddenly were using a lot of upload. Many homes still struggle with this today, and too many people upgrade to faster download speeds, hoping to solve the problem. ISPs using technologies other than fiber rarely mention upload speed.
Oversubscription. Home broadband connections are served by technologies that share bandwidth across multiple customers. Your ISP is very unlikely to tell you the number of people sharing your node or the amount of bandwidth feeding your node. The FCC’s broadband labels require ISPs to disclose their network practices, but nobody tells you statistics like this that would help you compare the ISPs competing for your business. The cable industry ran afoul of this issue fifteen years ago when large numbers of homes began streaming video, and many ran into it again during the pandemic. It still happens today any time a neighborhood has more demand than the bandwidth being supplied.
Latency. The simple description of latency is the delay in getting the packets to your home for something sent over the Internet. Latency increases any time that packets have to be resent and pile up. If enough packets get backlogged, latency can make it difficult or impossible to maintain a real-time connection. Latency issues are behind a lot of the problems that people have with Zoom or Teams calls – yet most folks assume the problem is not having fast enough speed.
Prioritization. A new problem for some broadband customers is prioritization. Customers buying FWA cellular wireless are told upfront that their usage might be slowed if there is too much cellular demand at a tower. Cellular carriers clearly (and rightfully) give priority to cell phones users over home broadband. Starlink customers who buy mobile broadband are given the same warning. Starlink will prioritize normal customers in an area over campers and hikers. Most ISPs say they don’t prioritize, but as AI is introduced into networks it will be a lot easier for them to do so. Over the last few months I’ve seen that several big ISPs are considering selling a priority (and more expensive) connection to gamers at the expense of everybody else.
Your Home Network. Everybody wants to blame the ISP when they have problems. However, a large percentage of broadband problems come from WiFi inside the home. People keep outdated and obsolete WiFi routers that are undersized for their bandwidth. Customers try to reach an entire home from a single WiFi device. Even when customers use WiFi extenders and mesh networks to reach more of the home, they often deploy the devices poorly. If you are having any broadband problems, give yourself a present and buy a new WiFi router.
Reliability. If operated properly, fiber networks tend to be the most reliable. But there are exceptions, and it all boils down to the quality of your local ISP as it does to the technology. It’s hard to say that any factor is more important than reliability if your ISP regularly has network outages when you want to use broadband.
I’d like to see the data that says fiber networks are more reliable. I’d argue that it’s false. There’s nothing inherently more reliable about fiber vs DSL or cable, fiber is the least resilient ‘wire’ to run to a home. wireless can suffer from noise but also has no fear of a backhoe or a car hitting a pole. fiber is also more costly and time consuming to repair, so outages are often much longer.
This seems to be an assumption made by those promoting fiber with zero evidence.
fiber has 2 noteworthy traits and that’s capacity now and in future, and low distance losses. That’s essentially it.
I would 100% agree with Daniel here. I want to see numbers that prove fiber is more reliable.
I know it’s not comparing the same thing but we have multiple fiber upstreams servicing our Wisp and I can assure you, we have more down time on the upstream fiber than any piece of our wireless network.
In 2024 we logged 62.25 hours of down time on our fiber circuits and in 2025 we are already up to 27.5 hours of downtime.
We do like any solid ISP does and purchase multiple upstream connections so that we can keep our clients online during the hours that one of the fibers is down. Very few of our last mile connections see more than a couple hours of downtime in a year.
Trendaltows, if you’re getting less than “two nines” reliability from your fiber-based carrier then I recommend two things to seriously consider: 1) Change provider. 2) Run your own fiber. *No* fiber network should be so unreliable.
I don’t see those as having anything to do with the physical medium. most of the businesses in the first paper have multiple delivery methods.
PDC 1) changing tier1 providers is simply not an option for MOST because the government has make virtual monopolies. 2) at >$100k/mile plus ongoing maintenance times hundreds of miles, doesn’t seem like a way to bring costs down.
The point is that it’s not the fiber interface (or copper or rf) it’s the company that runs it that makes it reliable or not. Tier1 providers have frequent outages. I had 7 separate lumen circuits go down yesterday. 2 zayo yesterday. comcast, wow maybe a dozen >1 minute outages YESTERDAY. I had more DIA fiber drops than I did consumer rf drops in a week and most of the consumer stuff is people’s own power.
I’ll double down here, there medium has little to do with the reliability, the company the runs it does.
@PDC your comments are not rational. We’re a local fixed wireless company. Are we supposed to jump into the long haul fiber market to try and improve on the uptime of AT&T and Comcast? We have multiple transports that are over 200 miles long connecting us to a DC. Those are the circuits that have the high downtime. From AT&T.
In regards to your other comment, we already have 2x 10G and 2x 1G from AT&T and a single 1G from Comcast. There isn’t any other fiber available in the area. We have been waiting for months/years on Comcast to offer more than 1G but they say they are dealing with a congestion problem to our whole area (Northern California) and they can’t deliver more than 1G.
Both fiber and wireless break. Fiber breaks are not as often, but take much longer to repair. The majority of outages on our fixed wireless network are related to some equipment malfunction that is 80% of the time fixed within 15 minutes from a remote location.
It is our experience that end users are much more tolerant of a 15 minute outage than a full day outage. And the outages that occur on our fixed wireless network are each one a learning experience with an improvement usually made to prevent that problem from happening again. Radios are cheap, redundant paths are easy and quick. I can deploy a 1G link out to 5 miles in a couple hours with supplies on hand. Fiber outages on the other hand don’t have much of an “improvement” aspect to them. Just dig it up, repair the break, and hope that the next construction crew misses with their backhoe or that the next vehicle accident doesn’t take out a pedestal/pole/etc.
I’m on numerous fiber and wireless group discussions and I don’t see any less outages on fiber than I do on wireless but that is certainly not an in depth study, just general observations that could be biased.
Paper based on FCC MBA data: https://arxiv.org/abs/1709.09349
Another one, from 2022: https://ieeexplore.ieee.org/abstract/document/9829788
Do you have data-based evidence that contradicts these?
DSL reliability is hard to assess, given the often miserable performance. But for HFC, more active components in the outside plant and noise issues seem like possible explanations.
Anecdotally, my experience with fiber hasn’t been marred by any outages so far, while my cable modem service used to crap out at least once a week. The typical outage didn’t last long, typically five minutes or less, so it would be hard for a uptime surveillance to detect. But it was annoying nonetheless. My wife, who isn’t technical, remarked on the reliability of the fiber service without prompting.
Surely, someone has studied this question by now.
@Richard Bennett, I would suggest that it’s probably new vs old not fiber vs coax. nothing about copper wires causes them to shut off or lock up occasionally. shiney new ONU vs janky old/recycled cable modem.
Agreed! Jason Livingood has a great quote about speed: “Bandwidth has been a great proxy for improving QoE for many years but we are now at a point of diminishing marginal utility of more bandwidth. To wit: the only app that needs 1 Gbps is a synthetic speed test. ;-)”
Do you know about LibreQoS? It can make a huge improvement on latency, and more importantly, latency under load.
Sure do, we have half a dozen LibreQoS boxes out there in one capacity or another.
to be fair to ‘speed’ here, there are occasions that I really enjoy pulling those gig archives down quickly. It’s rare, but it’s nice to have. And if I were just a customer of an ISP I would buy the fancy plan, I don’t expect that for $50.
As a person that pays for all of the elderly inlaw’s services, I would buy them a 50Mbps for $50 plan every single day and never think twice about it reliably delivering 30Mbps down and about 7Mbps up so they can have smooth facetime calls with family.
I’ve said it before, I’ll say it again. All we need is a healthy competative environment and if there’s a gap in the market that prevents that, a gap that the government could fill, it’s access to long haul (often existing) fiber to internet exchanges.
The speed argument is meant to eliminate all but 1 vendor in a marketplace. It’s the model the government designs and endorses and it has wide consequences in quality and service.
Meta comment, for Doug:
Have you thought about posting to LinkedIn as well as your blog?
Ha. I do post on LinkedIn every day!
Where the emoji for “deeply embarrassed”. 🙂