Unfortunately while 192.168.100.1 as a diagnostic web interface was reliable for many years, more recently a lot of cable companies have been locking down even these read-only interfaces preventing users from accessing useful data about signal strength.
A few years ago I had a similar problem with Armstrong Cable in NE Ohio and went through a similar process that involved multiple unhelpful tech visits eventually culminating in me writing a scraper to read the diagnostic page and collect SNR/Tx power/BER stats per channel plus some investigation with a SDR to prove that the two channels having severe problems overlapped with one of the frequencies used by a nearby AT&T tower, and that I never had problems when I wasn't attached to one of those channels.
After all that they did some kind of RF leakage test and found that chipmunks had been chewing on the RG6 line from the street so that got replaced with the big thick orange cable and all was well for the rest of the time I lived there.
About three months later they pushed an update to their modems that changed the seed used for generating the random password that "protects" the advanced diagnostics page from the default so neither I nor anyone else on their network could access it anymore without inside information.
They went out of their way to take objectively useful diagnostic information away from us. A pure evil move, but of course one they could do with impunity because most of Armstrong's service area their only competition is Frontier's ex-Verizon DSL which was hot garbage 20 years ago and has only deteriorated since then. No one who would ever care about that feature is going to change providers over it going away, the sad state of the US broadband market means they have no real choice.
Cable ISP here locally has gone as far as to entirely remove access to the 196.168.100.1 diagnostic page. I can't even tell if my modem is properly synced with the network or not now. :-/
> Can't you buy your own cable modem and use that?
Unfortunately no, Armstrong did not support bringing your own modem. It could technically be done prior to DOCSIS 2.0 but it was never officially allowed.
In the US providers are not required to allow users to own their modem, but they have to provide a modem for free if you can't bring your own.
This is mostly fine as long as they offer a plain modem, it's when they don't offer a model without its own (almost always absolute garbage) firewall/router built in that it's a problem. IMO it shouldn't be allowed to call something "internet service" unless I have a handoff I can plug in to a firewall of my choice and have a publicly routable internet address directly on my box's WAN port
> it's when they don't offer a model without its own (almost always absolute garbage) firewall/router built in that it's a problem.
That is a problem, but you can work around all of that. All of the ones I've seen have allowed you to just turn that stuff off, but even if you can't, you can approximate it by setting router rules that just forward everything to your real router.
> All of the ones I've seen have allowed you to just turn that stuff off, but even if you can't, you can approximate it by setting router rules that just forward everything to your real router.
The BGW320 gateway that AT&T provides with its 2g/2g and 5g/5g fiber services has a NAT table limited to 8192 tracked connections. Let me repeat that, a 5 gigabit symmetrical XGPON service has a NAT table that makes a 2004-era garbage tier device look good by comparison. It does not support a bridge mode, only "IP Passthrough" or the standard default NAT "DMZ" mode commonly seen on consumer hardware. Both of those rely on the tiny NAT table, so anyone who wants to establish more than 8192 connections at one time has to get a routed block of static IPs that allows the NAT to be entirely bypassed.
At least officially, for now there's also a method to clone your ONT identifiers on to one that works as a plain bridge, which I have the parts for but haven't installed it yet.
Their high speed DSL "U-Verse" services have a similar issue with a bad mandatory device, those require a complicated bypass involving passing through 802.1x packets to the provided router so it can authenticate while you then use your own to actually send/receive internet traffic.
I keep seeing references to this from Americans, but is this actually a thing? There’s no way my ISP would allow me to connect a device that isn’t owned by them.
When Americans complain about Comcast, it's not a meme, they are actually an awful company.
But for the record:
If you just sign up for Comcast (XFinity, whatever) and use the internet, you are renting a modem/router combo from them for about $10 a month, in perpetuity. You can buy an off the shelf modem/router for a couple hundred bucks, and use that instead. It'll pay off in a year or two. You have to call Comcast up and activate the device, but it's otherwise a completely painless process.
Further, my experience with Comcast specifically is that every three months, they "forget" that I don't rent a modem from them and just start charging me the rental fee anyway. Then I have to call them up to tell them to remove the charges and stop adding new ones.
Here you don’t pay a fee for the modem but they don’t let you run your own. Among other reasons because they want to force retire unsupported frequencies.
EU rules do not necessarily demand it - it depends on how exactly things are structured logically. They do demand that what you connect to termination point of ISP network is your decision, but that termination point can be a cable modem for example.
Shades of pre-1968 AT&T. "Oh no, you can't connect that modem to our lines." In fact, prior to late 1956, AT&T even told people they couldn't mechanically attach a device to their phones.
I at one point spent a hundred bucks and bought a used JDSU DSAM 3300 cable testing meter. It's only rated for DOCSIS 3, but I can get some good info out of it about my local cable plant and figure out if it's my side or the outside plant.
Then they show up and use their shiny new Viavi (formerly JDSU) OneExpert... and see the same thing and fix their shit. ;)
I wish they get some telemetry back from the modems and had some kind of KPI-driven process where they must proactively go out and rectify this stuff.
In Australia a lot of people think that the new NBN (national broadband network) is terrible, but it is actually one of the most stable and cost-effective broadband networks. The reason people think that it is terrible is because older houses are connected via telephone cables that have a whole bunch of issues including poor workmanship left over from the previous private owner, whereas new estates that connect directly to the NBN via brand new fiber optic runs are incredibly fast and reliable for a consumer service.
As a matter of fact, the NBN at my house costs around $100/month on fiber whereas my office/company business Telstra fiber costs around $1000/month. The speeds are comparable, but the SLA is different. Nevertheless, they have both been down about the same amount of time in the last year (not more than a few hours total). My mate who works at NBN thinks that this is because the government was allowed to operate NBN backbone at a total hemorrhaging loss of money (so expensive equipment and contractors are used), whereas Telstra has to operate profitably.
I lived on a street where we discovered through Nextdoor that we were all suffering from cable modem outages at the same times. It persisted for six months before whatever piece of equipment that serviced our street died for good. Only then did we get it fixed.
It's absurd that an ISP isn't monitoring, correlating, and responding to telemetry to inform of these problems without a trouble ticket.
Our neighborhood would periodically have 200 foot electric extension cables, running around yards and such to power the equipment. They would be in place for months before they eventually got new powerlines dug. Happened multiple times over the years, so it was quite common to see orange lions running all over the neighborhood.
> I wish they get some telemetry back from the modems and had some kind of KPI-driven process
Another lifetime ago back when cable modems were still fairly new (and uncappable!) you could do this yourself. The modems would all be on RFC1918 space, and be completely reachable to anyone who thought to look. They had a standard SNMP MIB which you could use to graph out signal strength, bandwidth usage, etc. I of course fired up rrdtool and started mapping out my neighborhood.
It was quite interesting to be able to detect problems and actually respond to them proactively. Since the modems were all default settings, you also had r/w SNMP access so you could write to the "reboot" OID among others. It was pretty trivial to script something that would boot off the habitual heavy users during peak times when you wanted to do some gaming.
Fun times. But I also always wondered back then why the hell Comcast (or whomever) wouldn't pay some kid like me $15/hr to build the same telemetry.
I wish I could get the telemetry, so I know when the modem wasn't working and needed a restart, or the NBN is just down...
I'm not what networks you're comparing the NBN with, but NZ is far better...
Also, not sure where you're getting the ~$1000/m number from, Telstra 's SMB NBN rates according to their website are also in the $100 range (no numbers for enterprise naturally), ~$1000 is probably either private Telstra fibre and covering the install cost or has better bandwidth/latency properties (from the NBNCo website it looks like there's Traffic Class 2 which is symmetric upload/download speeds, and Traffic Class 4 which is asymmetric, with Traffic Class 4 sounding the same as the home plan).
I get an alert for an internet outage (I use Grafana Cloud so the outage isn't preventing me from being alerted). I also have the side benefit of historical information on the health of the connection.
Years ago at the house we were renting our cable Internet would have intermittently go down. It would only last a couple of minutes and only happened in the mornings when the sun was coming up.
I was convinced that somehow the line from our house to the pole in the alley was bad and when the sun was warming it up it caused the issues. The cable company refused to fix it since they never saw a problem on their end by the time I would reach customer support.
My solution was to upgrade to business cable Internet, which was about the same price we were already paying, just required signing a 1 year contract.
First day the new service was active, called business support and 2 days later they came out and replaced the line. No issues after that.
I too went with the 'business class' cable a couple of years ago. In my market, it's dollar-for-dollar slower than the consumer grade (I get 250Mbps where I'd get probably the 500Mbps consumer package for the same price), but in exchange I get an actual SLA that they mostly adhere to. Given my WFH gig, it's worth every penny to have an outage addressed (if not resolved completely) in 4 hours rather than 2-5 business days.
I was facing consistent multi-hour outages during the workday and looked into business internet options with my ISP (Cox) and from what I could gather they _did't even offer_ any plans with an SLA in my market.
Though I loathe Cox for several reasons, in my neck of the woods, they offer wifi with excellent coverage over the neighborhood. Captive portal can be logged into with Cox account credentials.
I've used it several times as a backup, e.g. during power outages. Not a bad perk.
I loathe Cox for many reasons mostly related to their unpredictable usage policies (perennial shenanigans with the word "unlimited"), adversarial customer support and messaging, and high prices with dark patterns re: bundling and promotions. (though honestly these complaints seem par for the course when it comes to American ISPs.)
Buuut at least in my neck of the woods they offer wifi with excellent coverage over the neighborhood that you can sign into with your Cox account. I've used it as a backup (eg during power outages) several times. Not a bad perk.
Well, yes Cox objectively sucks. But to your point, someone offers internet connectivity in your area with an SLA; the question is at what cost and how much a backhoe might be involved.
Last summer I had consistent "micro drops" in my upstream cable connection with a frequency of a couple times per hour up to dozens of times. The peaks were generally weekday afternoons.
The symptom was the same each time, which was a roughly 5-10 second period where my download would work perfectly, but no data could be uploaded. This was frustrating at work and at play. For the former, I'd see and hear everyone in a virtual meeting, but they'd lose my audio and video during that time. At play, I'd just get dropped from whatever online game I was playing.
Looking at the modem's status screen, I could see that I only had a single upstream channel connected. Looking to bypass wiring, I connected my modem directly to where the home wiring connects to the wire from the street. Same result.
When I called Xfinity, they let me know it was probably inside wiring, even after I let them know about the test at the wire entry point. They really didn't want to send someone, and warned me it'd be $150 if it was found that the issue was inside wiring. The next available tech appointment was 13 days away.
The good news was that FIOS was available and they could come the next day. Even better, I was able to convince the tech to run the new cable to a much more convenient location for my desk & router.
> Looking at the modem's status screen, I could see that I only had a single upstream channel connected.
Is this really a reliable indicator of a problem? I'm also on Xfinity, and see the same thing - 3 out of 4 upstream bonded channels are in the "not locked" state according to my modem. They've been that way ever since I moved here.
However, I receive the full upstream bandwidth that Xfinity allocates for my plan (a pathetic 20 Mbps), and it has been extremely reliable, no outages. Since it looked like this from day 1, I made the assumption that Xfinity only uses one of the channels since the bandwidth they provide is so low. The power levels and SNRs for all locked channels (up and down) are green across the board.
> Is this really a reliable indicator of a problem?
Apparently it was a problem (according to a tech I spoke with over chat) because having a single connection meant that any interruption in that single channel meant the entire connection goes down. Depending on DOCSIS version that channel does support the full bandwidth.
My upstream was generally around 10Mbps, which worked fine for nearly everything we did, but when things were bad, it'd drop to 1-2kbps or lower.
Wtf kind of company is Xfinity running? Spectrum offered to replace my internal wiring for free. I just didn't want some guy running around in my attic, so I did it myself.
What made it even worse was that when internet was down, I thought I'd just watch TV, which I also had through Xfinity. Nope, even though I could see and hear the channel, the STB requires full connection to operate, so my show would be on in the background, but 80% of the screen would be taken by a dialog telling me the STB could tune channels.
As it turns out, EVERY SINGLE REMOTE CONTROL PRESS received by the STB is then round-tripped to an Xfinity server, so when the internet has an issue, you can't watch TV. That made ditching cable and moving entirely to streaming a very easy choice since it killed cable's main advantage.
Anybody have a write up like this but for Fiber? Like FttH with calix 716ge-i?
I'd love to get into it's interface and diagnose things but it's complicated. It's on a separate VLAN when I connect it to any router and I can never actually figure out if I can and if it's actually possible to get to any sort of diagnostics from my side of the connection. Even with my overkill pfsense 9900k router with 2 Intel 4 port cards and wifi... I had to do some weird shit to get that to work when instead when I had my ASUS router it just worked.
But now I get way lower at times than my advertised speed and I really don't want to have a technician come down. I mean, I will eventually, but I want to tinker more first.
> Anybody have a write up like this but for Fiber? Like FttH with calix 716ge-i?
Most of the faults with Cable Broadband are caused by the unreliability of coaxial cabling - as the article shows, many of these can be solved (or at least diagnosed) in a DIY manner.
By not using Coaxial cables, most of those issues should be eliminated with FTTP connections, and that's why FTTP is generally seen as the most reliable type of home internet connection.
You're saying that a cable and connector designed in the 1950s to send VHF NTSC signals a few hundred feet to houses in suburbia as a replacement for rabbit ear antennas isn't sufficient for broadband internet in the year 2023?
It's funny, isn't it? Coaxial is so finicky. They have to adjust the signal levels year-round because temperatures affect attenuation.
Then they keep coming up with new DOCSIS protocols to stay relevant and somehow manage. The latest can do full-duplex 10gb. I'm still gonna switch to local FTTP at the drop of a hat if it becomes available.
Cable internet is the worst one I've had; DSL, FTTx, Ethernet, all have worked better for me than cable. One big problem is that the DOCSIS spec is region specific (EuroDOCSIS vs US) and the vendor interoperability is poor, so buying your own modem is impractical in many cases. And then you have stuff like some chipsets (hello Intel Puma) being just broken and still remain in the market.
It's like somebody was spying on me. Weeks ago, I had intermittent internet issues. Turned out the problem was 1) corroded connectors in house, 2) no ground block at the house (!!!), 3) cable to the distribution box not secured. Lineman fixed all three, signal became great.
It's interesting that issue 11 is written as a warning. It's exactly the solution I chose. My cable modem always switches to higher frequencies after enough time, which never have enough SNR to work properly. I can't really force the ISP to improve their infrastructure. But this only happens after around 36 hours. I have a script that power cycles the modem each morning when I'm asleep.
I've even gone as far at this point to actively query the status of the modem and detect when it moves to one of these channels. If this happens during the day the modem is immediately reset.
What happened in my case was that I had very legacy broadband from when broadband was just becoming available and which had gone through 2 or 3 providers since then. Cable TV from another company. That plus phone service replacing POTS eventually got rolled into a Comcast triple-play. Then I dropped the voice (with separate voice modem) and cable TV.
At some point my Internet just got unacceptably flaky.
Turned out the problem was that with all the changes over the years, there were just too many splitters and unused cable runs. Essentially the whole in-house wiring to the cable modem needed to be redone.
I feel like you can simply do a visual inspection of cables before running an entirely new line to your house as described in step 2. I had intermittent outage issues with my cable internet before, and was able to find where squirrels had chewed through the insulation on the outside line, which would cause outages or issues when it was wet.
My internet works luckily, but recently noticed that whoever installed the coax in the house we're renting just cut through one of the crawlspace vents, leaving a giant hole for mice/etc to get in through. Real quality work there...
A few years ago I had a similar problem with Armstrong Cable in NE Ohio and went through a similar process that involved multiple unhelpful tech visits eventually culminating in me writing a scraper to read the diagnostic page and collect SNR/Tx power/BER stats per channel plus some investigation with a SDR to prove that the two channels having severe problems overlapped with one of the frequencies used by a nearby AT&T tower, and that I never had problems when I wasn't attached to one of those channels.
After all that they did some kind of RF leakage test and found that chipmunks had been chewing on the RG6 line from the street so that got replaced with the big thick orange cable and all was well for the rest of the time I lived there.
About three months later they pushed an update to their modems that changed the seed used for generating the random password that "protects" the advanced diagnostics page from the default so neither I nor anyone else on their network could access it anymore without inside information.
They went out of their way to take objectively useful diagnostic information away from us. A pure evil move, but of course one they could do with impunity because most of Armstrong's service area their only competition is Frontier's ex-Verizon DSL which was hot garbage 20 years ago and has only deteriorated since then. No one who would ever care about that feature is going to change providers over it going away, the sad state of the US broadband market means they have no real choice.