Below is my survey response to their support feedback request:
1) The products should not have a built-in time-bomb that causes them to stop working after only a few years.
2) Dell should have informed us of this issue by email. Instead they left a very "phishing" like voicemail on a manager's phone. (Not the phone of the registered point-of-contract for the cluster.) Perhaps this was done to avoid leaving evidence of #1 above?
3) I spent over an hour on hold when I returned the call, and was then disconnected. After trying again (to an extension other than the one given in the message), I reached somebody who confirmed the issue. I spent another four hours on the phone resolving it.
4) Shortly after all of the above, I discovered a new issue that severely impacted the cluster. The n3048 switch would no longer auto-negotiate a 100Mbps Ethernet link. Our network watchdog device (iBoot) was continuously cycling the power on our Internet Ingress (ONT+ASA).
5) I spent even more time troubleshooting and resolving this issue (by locking the iBoot port to 100Mbps instead of leaving it on Auto).
6) I did not waste any more of my time by reporting this issue. The technician I worked with to upgrade these switches assured me that the firmware releases we used were "stable".
The non POE N1100's are fanless, Thankfully don't really contain any features that would require licencing, that being said also has no hardware Layer 3 capability, so not really in the same class as the N3XXX or N2XXX's
The licencing thing does suck though, that's poor from Dell who normally (at least switches wise) do a reasonably good job for the price.
First Turtle Beach bricked my Audiotron by abandoning the web site required for it to function, then Grace Digital bricked my three GD streaming devices for the same reason.
If you are referring to a Reciva based internet radio the web site which powers them was shut down by the owner of that site, not by Grace Digital. Also, if it is similar to my CC Wifi Reciva based radio it is still possible to access existing presets on the radio although you can no longer add presets or modify them any more. When you power it on wait for the Network Error caused by the lack of the back end and then press Back twice and then a preset. The media server functionality also still works.
Very interesting read and lots of upside to what is discussed, but the thought of the uncountable, almost invisible operating systems running in a large network give me a odd feeling in the pit of my stomach. So many potential places for malfeasance to hide.
Honorable mention goes to:
- the Dell switch with the changelog note "if you had a device with a MAC starting with these bytes plugged in, the last N ports on the switch will stop working"
[1] - https://www.dkriesel.com/en/blog/2013/0802_xerox-workcentres...
Watch it happen anyway. Watch it accelerate. Watch the devices grow in complexity, capability, connectivity, and vulnerability. Watch innumerable manufacturers go out of business with no software-update succession plan, no code escrow, no upgrade path for victi^H^H^H^H^H end users.
The pit of your stomach gets pretty damn sick of the state of things.
It's an entertaining and informative read, but it's more like low-end datacentre hardware than that cheap €25 switch I've got in my home office.
A little disappointing as I was hoping I'd have a cheeky high-bandwidth raspberry-pi alternative on my hands ...
Also current price on ebay for a refurb N1108T-ON unit is 360 quid, so I reckon relatively speaking you could say it's cheap.
It's even a very cheap fully managed switch.
N1108T-ON:
1GbE Port Attributes Multi-speed:
8x 10/100/1000Mbps half/full duplex RJ45 ports
1GbE Port Attributes Single-speed:
2x 1000Mbps half/full duplex RJ45 ports
Integrated 1GbE SFP dedicated ports: 2
Integrated 10GbE SFP+ dedicated ports: N/A"powerline adaptors" [0]
Basically, buy these adaptors and for ~$100, you can have create a "wired" LAN connection using power outlets in a home.
[0] https://www.techradar.com/news/the-best-powerline-adaptors
It took forever to figure out what was causing this, and I eventually figured it out by doing a (very slow) IP scan of every device on the network I was connecting to and finding a machine named with their first and last name. Unplugged the thing, and the problem went away forever.
If it hadn't happened to me, it's something I would have thought impossible!
(edited a small typo)
Speeds won't be great for that sort of ethernet-over-phone wire, especially because most landlines used awful, cheeap wires, but in some cases it is faster than powerline adaptors.
Unfortunately all the companies that produce such equipment for "phone line adaptors" sell only to the phone companies and never directly to consumers.
So many houses today have vestigial phone wires that no one is using for anything.
I used it in home automation (Thank you Home Assistant) for mostly sensors, and less important actuators, where I had power, but could not get Ethernet, or WiFi.
Anything else, very unstable and too much latency.
My apartment has some places that have coax runs but not ethernet runs.
Their performance depends on the electrical wiring in your house.
Ironically for me I’ve only managed to make them work on older electrical installations, and even then it was at best 802.11n speeds.
In my current house (newly built 2018), I can’t even get them to handshake. I suspect it’s due to improved electrical standards and better isolation between different circuits internally in the house, but I honestly don’t know.
> did the author drop a zero in the price?
Not far off ...
It's a fantastic device if all you need is 2 ports.
I originally had two mikrotik rb2011* switches and a rb750gl, now I have two rb3011 switches.
Linux has recently got native framework for control of switching chips called "Distributed Switch Architecture." This turns a lot of very cheap hardware with very basic hardware switching chips into high performance routers.
EDIT: No, actually, I was mistaken - there is one Broadcom series supported now I look again, BCM53xx, of which the one in the article does indeed appear to be. Looks like since the 4.8 kernel though, so not in the 4.4 kernel that comes with the device.