Hacker News Reader

Comment by nine_k | Hacker News Reader

What kind of access should a cable company have to your cable modem? Should it at all?

I mean, your ISP does not need any access to your edge router if the ISP gives you a standard Ethernet socket. How standardized are cable interfaces? What kind of custom setup may they legitimately need to work in a particular cable network?

View on HN

It makes a little more sense once you look at the infrastructure that it's running on. You and ~500 of your neighbors are all on one local node that shares bandwidth between everyone. The way it works for upload is that your cable modem sends a request for an upload timeslot on a shared upload channel, then the CMTS which serves thousands upon thousands of people sends back your time slot when your modem is clear to use the upload channel. If anything goes screwy on your node such as significant interference or if your cable modem somehow hung up and was constantly sending on the upload channel outside of its time slot this means that absolutely everyone on the same node looses service. Not only can it affect a ton of customers all at once, your ISP doesn't have a good way of tracking down interference at a finer granularity than the node level which isn't a lot to go off of. They literally have to do a binary search by disconnecting sections of the local node and seeing what segment the interference is coming from. I've heard horror stories about interference coming from things like washing machines, treadmills, etc and being very intermittent and nearly impossible to find because of it. For the treadmill story, a whole node of customers would loose service for 15 minutes every couple of nights and whenever the ISP would get a technician out to start trying to isolate it, they'd already be off the treadmill and the problem would be over. Supposedly it took a while of having a tech in the area around when it normally started to track down which house it was coming from.

So yeah, it's your modem, but they won't let you use their network unless you're using one of their approved modems that they know are reliable and that they can manage updates on because just one person can unknowingly screw it over for a large chunk of people.

The cable modem is telco infrastructure that happens to be in your house. The boundary between telco and customer networks (called the "demarc") is between the cable modem and your router. It's entirely theirs to administrator, same as the vault down the street.

A comment I posted about this a few years ago: https://news.ycombinator.com/item?id=6998650

It's not quite that simple. In some cases it's telco infrastructure that's owned by the customer. In that case I'd say the demarc is like to be somewhere inside the modem, which is sort of nonsensical.

When you own the modem, you're free to disconnect it from the cable company's network and do something else with it, but administrative access is still a condition of service from your ISP as long as you continue to contract with them for internet access on it.

In the case of cable internet service (not fiber), you'll almost never get a standard Ethernet socket. Chances are, you'll get a coaxial connection that requires a device to bridge the connectors and perform a handshake with your upstream provider.

Of course, most consumers go with whatever hardware their provider gives them (usually a gateway to provide Wifi). This presents it's own problem: in the US, cable companies are trying to set up mesh networks/guest access, and so those gateways may be running a second semi-public as a node on the mesh.

We just got new tenants in our (commercial) building. I'm honestly a bit upset as there are now _four_ new access points for a single tenant.

"TWC WiFi", and "CableWifi", both unsecured (!!!), and then "TWC WiFi Passpoint" (which requires a TWC subscription to use.)

I sure wish people wouldn't blindly trust the cable technician to configure their wireless network properly. Now there's just tons of RF noise, and people can leech bandwidth off our building. -- I frankly find it ridiculous, given the premium we pay for commercial internet (which is slower than my residential subscription), that we are expected to share it with their "mesh network."

Aren't those SSIDs all on the same WiFi channel? Having split guest/internal networks broadcast from the same radio on the same channel really doesn't hurt things enough to care about. And it's almost certainly using separate DOCSIS channels or not counting toward the traffic shaping limit on your traffic, and it's prioritized lower than your business-class traffic when it's further upstream in TWC's network, so you don't need to worry about it affecting your WAN connection performance either.

The fact that the telco/cableco broadcasts a public or subscriber wifi off an installed business class service line is more WTF than potentially misconfigured broadcast channels. This is partially why I bought a SB6141 for my home. It has no radio. Maybe I need to go shave my neck, but I'd rather have a device that does one thing well than one thing that tries to do everything adequately.

Given that the business is merely renting the modem/router/AP device from the ISP, the only theoretical downside to the business is the extra electricity used by that functionality. And for quite a few businesses, the zero-effort zero-liability hotspot for customers is a pretty big upside.

If you want thorough control over your network, then of course you won't rent a modem from the ISP and you'll install a separate router of your choice and decide for yourself whether to operate a hotspot. But for the majority of customers, none of that is worth thinking about and the ISPs are actually providing sensible defaults.

Reply on news.ycombinator.com