After reading this thread I thought I would post how not to implement Lede/OpenWRT in a consumer product.
I was reading a review of a yet to be released router made to make it easy for the user to connect to the internet using a VPN connection, Some of the mistakes made in the configuration of this router are scary as far as security are concerned.
Vilfo VPN router review:
Overview and hardware – Part 1/4
Vilfo is an upcoming VPN client gateway intended for home users to shield their activities from their internet service provider. It’s much more powerful (and expensive) than your average VPN capable home router, but is powerful hardware all it takes to make a good VPN gateway?
This review is divided into four parts.
Vilfo is a powerful 430 USD network router using mini-PC hardware rather than the traditional barely-enough processing power you can expect from a traditional router. Vilfo envisions you’ll use this extra processing power to encrypt all your traffic and tunnel it through to a remote virtual private network (VPN) provider.
With regard to ath10k and air-time fairness: The firmware itself tries to do ATF, at least with the wave-2 (10.4) firmware. Higher-levels in the stack could do some fairness with current API AFAIK, but maybe not quite as precise as ath9k. And, in my own testing, I had to disable ATF in ath9k in the 4.13 kernel because it has (hard to reproduce in a sane environment) tx lockup bugs. All in all, I think ATF is not a huge deal one way or the other...stability and basic features is what you should look for. Ath10k is not perfect on that, but I don't know of any better /AC chipset and it works pretty well for us, at least.
Thanks for your response, one of the really great things about this forum is that so many experts like you are willing to share their knowledge. So the race for the moral successor to atk9K is still not done, even though ath10K has quite a headstart
I guess I really really need to start playing more with my turris omnia, since it sports an ath10k and is "supported" by stock openwrt since recently.
I'd like to get back to this topic. The issue here is not about commercializing OpenWRT. There are always good and bad efforts and the goal here is not to criticize those. The issue is with the availability of ideal and low-cost hardware for prolonged time periods. Why does the answer to the question users frequently pose on ideal low-cost hardware have to constantly shift? Does it make sense for OpenWRT to control more of this process? Is the size of the community enough to lay out reference architectures and have vendors subscribe to them? I think it's a secondary effect that if hardware is more planned and less unpredictable, there is a ton of value added in OpenWRT that might trigger commercial interest.
I respectfully disagree. Despite the bundle of software and hardware - I have bought from AVM in the past - it is basically hardware since that's where their customers have been asking in their requirements. The software is there just to support the hardware performance and security. The firmware upgrades are a good sign of a responsible company but that does not change the box-moving nature of the market and the imprinted DNA of AVM and all the vendors. Btw I hate to be citing AVM on this, they're one of the greats but such is the nature of this market.
ideal stuff shifts because the prices of chips fluctuate a lot and depend in large part on the manufacturing volume and the sales volume. What people are using a lot of today is what is cheap, and vice-versa.
No, it doesn't make sense for OpenWRT to "control more of this process" because this process is a global economic supply and demand process.
No, the size of the community is nowhere near big enough. TP-Link is apparently more than 50% of the worldwide router market. That market is for hundreds of millions or a billion units a year.
There is a limited number of devices on the market at one time. Development should be aimed at them where possible. Hopefully new releases will continue to work and when they don't someone with interest will fix it.
This project is useful for the latest devices as well as recycling end of life devices probably more the end of life devices. Because by the time a device is fully supported it is usually nearing end of life anyway.
The contributors on the project have barely enough time merge LEDE and openWRT and now people are asking for project hardware as well.
If you have a business idea go for it. Don't put the burden of your business on to the project. Design or buy your hardware. Donate hardware if you like, do your own development your self or with your team and push back the improvements so the community can continue your work just in case you go belly up.
How long before this topic is solved. There is no ideal specs is the simple reason. Everyone has different needs and expectations. Just make a choice and go with it. If it works bloody great, if not try again.
But this is what customers want to buy: real tangible products; I guess this is the same reason why you try to create a reference platform, so that openwrt interested parties can be easily steered towards reasonable hardware, no?
That, I believe, is a decent description of a lay person's take on the matter.
Well, I dragged them in to show that there are commercial entities that seem to have some clue.
Obviously not supporting that, just saying that this is the current mindset in the box-moving, commoditized home router market and that's why you see limited firmware updates and support by a number of particularly low-cost vendors.
