I mean this is pretty standard in all industries regardless of whether it's a software flaw or a physical flaw in any other kind of product. What's the likelihood of a vacuum manufacturer replacing a part in a 15 year old product that had a 1 year warrantee even if it's a safety issue? Sure the delivery and installation is cheaper with software, but the engineering and development isn't, especially if the environment for building it has to be recreated.
You can argue over what a reasonable EOL is, but all hardware is going to EOL at some point, and at that point, it isn't going to keep getting updates.
Throw enough money at a vendor, and I'm sure that you can get extended support contracts that will keep it going for however long people are willing to keep chucking money at a vendor -- some businesses pay for support on truly ancient hardware -- but this is a consumer broadband router. It's unlikely to make a lot of sense to do so on this -- the hardware isn't worth much, nor is it going to be terribly expensive to replace, and especially if you're using the wireless functionality, you probably want support for newer WiFi standards anyway that updated hardware will bring.
I do think that there's maybe a good argument that EOLing hardware should be handled in a better way. Like, maybe hardware should ship with an EOL sticker, so that someone can glance at hardware and see if it's "expired". Or maybe network hardware should have some sort of way of reporting EOL in response to a network query, so that someone can audit a network for EOLed hardware.
I think there should be a handoff procedure, or whatever you want to call it.
As EOL approaches, work with whatever open router OS maker is available (currently OpenWRT) to make sure it's supported, and configs migrate over nicely. Then drop one last update, designed to do a full OS replacement.
Definitely don’t this in the past (Linksys WRT54G!) but let’s be honest, the kind of people running 10yo Dlink routers aren’t going to flash new firmware, let alone OpenWRT or even know to look for it. It would have to come that way from the factory. And even then I doubt most people even do regular updates, sadly.
Something this old is going to be power inefficient compared to newer stuff, and simply not perform as well.
I would know, I just booted up a 10 year old consumer router last night, because the current one died. It'll be OK for a few days until I can get a replacement. Boy, is this thing slow.
When the users are in control of the software running on their devices then "EOL" is dependent the user community's willingness to work on it themselves.
Commodity hardware & open source software for the win.
When my Western Digital NAS was never going to get critical security patches, I was so freaking glad to find out that they just used software raid... I threw the HDDs in a Debian server and never looked back.
It's certainly nice to have things that are turn-key, but if you can find your way around any OS, just avoid proprietary everything.
Instead of trusting DLink with an off the shelf NAS, it might be easier to build your own with a Raspberry Pi running openmediavault hooked up to a couple of USB hard drives. It's worked well for me for over 6 years now with no issue and could cost way less.
I watched and enjoyed that one yesterday, and he's bang on the money. People here are saying "well it's EoL" but that means it's got all the way through development and its full lifetime with such a prominent set of bugs.
I don't think I'll be buying D-Link if that's what supported means.
Same website (granted, different author, but), same inflammatory language, same vendor, referencing previous erroneous article...I'm not even gonna read this one. Just going to copy/paste my previous response from the previous post:
At a certain point it's the consumer's (and blog writer's) fault, and that's after EoL. Not patching a supported one and just getting rid of support, saying buy a newer one? Yeah, that's bad.
Continuing to not support an EoL model that you already don't support due to EoL (or even dropping support for an EoL model that no one expected you to support in the first place due to EoL)? Non-issue.