Mesh + Dual Wan

[zaqhack]

After a couple of weeks researching this, it feels like a unicorn. Is there any way to buy/build this kind of thing?

 

[coxhaus]

Why do you need mesh? Mesh does not mean wireless.... only in the consumer world.

You can do this in the small business networking world without true mesh. If you want true mesh I think you need to move into the enterprise world.$$$$

 

[Trip]

@zaqhack - What is your overall goal here? If I had to guess, I presume you're trying to provide multi-node wireless for a home/SOHO network, whilst ensuring as much hardware and ISP service redundancy as possible upstream? If so, there are likely better ways to set this up than how your drawing might suggest, but you have to specifically define what you're trying to protect against, as it likely required differently-tailored engineering, from the LAN to WAN and beyond, some of which can likely be achieved, and some of which may not be able to be achieved.

For example, if your goal is to simple prevent against gateway hardware failure, you can use an active-passive, high-availability pair (with VRRP). For ISP failover, you can simply load-balance out of a single router. To mitigate both gateway and ISP failure at the same time, that could be doable, but depending on how you want it setup, would potentially require four total ISP links, two from each provider, with a pair connected to each router. I'd imagine cost is a factor, so it's important to be reasonable with what failure domains you're trying to control against, including making an accurate assessment of how often they're likely to fail, and whether or not that type and amount of failure is permissible to you, given your ultimate goals.

On the wifi side, a 3-node, multi-point mesh is achievable, even with certain consumer gear, but setting up the each root mesh node to backhaul traffic ultimately to different WANs is a possible recipe for disaster traffic-wise, and/or creates a single point of failure in and of a setup like that itself. Again, any setup here is potentially doable, but again, at what cost, and for what practical purpose?

We need to know more about exactly what you're trying to achieve, and why, to recommend the right solution for you.