hi there. I've been looking to setup a mesh network but before I get to any specific questions let me explain the scenario, maybe someone can give me some tips.
I'm trying to setup a mesh network in a small touristically developed village but technically lagging behind. So basically, the main road is a couple of hundred meters long with about 20 locations (restaurants, bars, shops) out of which maybe half have free wifi for their customers but they are badly set up, sometimes they drop connections, most of the times the guest wifi isn't separated from the LAN.
So my initial idea was that I could cover this whole area with about 6-8 APs and I'm sure 3-4 of them I could use as gateways by plugging them into the internet connection of willing participants while the other 3-4 would simply extend the range and act as mesh nodes.
But now I realized that say half of these people use the same internet provider and are probably using the same router at home with the same DHCP range while the others have different providers and different settings. I guess this won't work if I just plug my mesh APs in there.
So what is the solution? Ask to adjust their DHCP range? Have all routers give out a portion of the same range?
What is the way to go forward?
I was looking to go with ubiquiti hardware, as they have nice management software to go with their products.
Will APs for home usage be enough? Is there a slightly more advanced solution? I'm sure I could get approval from the village to set up something bigger provided the costs aren't exponentially higher.
On the other hand, mesh isn't 100% needed, the most important part is to be able to manage all APs centrally so should I use Ubiquiti APs instead of Mesh APs? That still gives me their management interface, I can still use vouchers for all APs, etc.
You will have no problems if you set up your own subnetwork, imagine that you have 6 Dumb AP, with lan interconnected, and 3 of them have Wan also connected, but in their DHCP it rotates itself the gatewate IP's, so when a client receives dhcp config it will point to one or another of those 3 LAN IP that have WAN connected, so they will browse the web using some or another...
Another way is to have a multi-wan setup on some router with the all gateways IP configured and all client connect to that router, but since you will using wifi as the mesh connection you will have double transference and will be really slow (even if you add another dedicated wifi link, only just for the link.
If you need this to grow easily you will need to set this same but using some mesh software on the AP's... and configure acoordingly to take care of the multiple gateway configuration.
You could use any hardware that you like, Ubiquiti and Microtik are known to be good enough for serious bussiness I think that Ruckus is also very good too. But if any of those hardware vendors does not have a mesh like able hardwire capable (since limitations of their firmwares) you will have to try using some other Hardware with LEDE,... but even this way if you connect Ubiquitis/microtik to the link acting as just Wifi links and AP and put the mesh logic of gateways in some device with LEDE.
If you dont need mesh, you could enable an additional ssid with ssh enabled, or enable ssh with portknocking on wan iface (you should take care of opening port on modem) ... or reverse ssh using ngrok like method, or even better, set up your own OpenVPN server on the could, and configure each lede router to connect an additional network for management to that VPN, you could also connect to that VPN to manage and inspect all the routers.
Good luck and if you had time keep us posted on the development of your project, it will be insteresting for us.
Will this network be open for anyone to connect (and also anyone to add a mesh point) or limited?
I don't know what all is out there. Lately I tried meshdesk but can't really recommend it. I have used OpenMesh and though it is closed source, it works very well. They really want you to buy their hardware. They do host a cloud controller that is free to use. Freifunkt is very long established and widely used in city-wide operation.
Are you only looking to fill in gaps in the tourist street, or expand to the surrounding neighborhoods? In the first case it is going to be perfectly adequate to set up some simple repeaters using a regular AP-STA connection to single-hop to the nearest ISP served location.
Well, the purpose of this little project is to monetize the wifi access. Currently wifi hotspots are few and badly setup creating more work for the businesses supplying them.
I need an easily manageable network so after looking at solutions, I decided to use Ubiquiti. Lets not talk about vendors yet, its mostly irrelevant.
Say I use Ubiquiti, with their central management, I can deploy configs and vouchers, etc. to all mesh APs so I can i.e. include some vouchers in our local tourist brochures, this way word gets out that there is a central wifi authority in town and new tourists can test it with the vouchers included and they can also buy more.
The idea is not for this network to be extended except by myself. I have a deal with some businesses, allowing me to use their internet, meaning I can plug my AP into their router and provide them with a certain number of vouchers for free and for sale as well as maintain my own devices.
So yes, I can plug my APs into their routers but if I configure these APs as a huge mesh IÄll run into problems due to DHCP.
I just returned from a week in Central America and definitely see this type of need in small villages.
I applaude your initiative and will agree that the Ubiquiti hardware is a great choice. I am not sure I agree with your implementaion. Separate the marketing (Vouchers) from the infrastructure. They are just funny money.
It looks like a hand full of the Unifi Mesh devices should cover the street, along with some additional hardware (router and AP-LR I think from quick reading) for you to build your own independant (under your control) WIFI network. It looks like the firmware has it's own portal tools which support payment, so credit cards and vouchers become different forms of payment.
Trying to integrate multiple intenet connections will be a lot of pain. Pave the street (line of site), so to speak, on your on (most control) or with help of the best (single) connection you can partner with (you are still dependant on thier connection). Then offer the other business some type of "Managed WIFI" which basically (from your perspective) means renting them a Ubiquiti Mesh node for inside their establishment. (What ever device you choose, buy and keep preconfigured spares on hand, remember you are a service.) Pick a number ($99) for a set up fee, which mostly covers your hardware and charge them some type of monthly fee to "Maintain" it ($19-29, pay off the hardware in 6 months). Connecting these should be a no brainer once the street is paved.
Users will be able to roam seamlessly (Sell the great experience). You want to keep the fee to the shops low so all will install the APs which strengthens your backbone. This revenue should cover your maintenance, but your real revenues come from the tourists at say $9.99/day pass. I think this is what the connection fee was for a day pass in Rotan Honduras when I was there. I believe the hourly fee was $7.99.
Back to the vouchers, consider offering the vendors some type of discount or commission for selling these. Not sure if the portal software can help you manage this. I do not think they will want to "Inventory" these passes, so some type of real time option may be needed.
let me just make sure I understood them right: you are basically saying, don't bother with the pain of trying to use multiple internet connections which are not under my control: either get them under my control then I can manage them or use them while not under my control by simply having them power my APs but in that case forget about mesh?
So how does this sound:
get 3 well placed centrally located internet connections under my control, acquire 3 Ubiquity routers + 3 APs and place them at these locations then add 3 mote APs or Mesh Aps to connect these 6 APs into a network. Try and sell it, then expand?
Last question: is this type of consumer hardware the right way to go or is there some pro hardware which is way more suited for this solution?
I am saying use 1 internet connection (router)* and multiple APs (2-4) in a mesh environment under your control (ie your internet connection\router). This gives you total management control.
If you use your AP off their network you are both subject to the partners network working (issues, security, QOS, etc) and probably need to be at that location to manage the AP. Also you may need multiple sets of tools to monetize things and expect this to be harder to control.
Expand from the street into the businesses by leasing them a node on your mesh network (obsoleting their need to maintain a public WLAN and the "headaches" that come with it).
I do not consider the ubiquiti hardware consumer, but I also expect that there is better. I think for under $1000 you can get everything you need to start (pave the street) that is well integrated and supported by a single vendor. Buying an integrated solution will be much easier to manage and support than a hodge podge of product (where the vendors can point fingers at each other). This will also let you focus on the business plan. If the business grows well the $1000 will look like a cheap investment
You probably want to move this discussion to the ubiquiti forum.
Or a single router that is designed to load balance multiple ISP connections, again under your control.
Right you are, lets start small and expand. The only reason I initially discarded this starting scenario was that I was scared of the loss in speed due to potential multiple hops (seeing that I am sure some APs won't be able to connect directly to the gateway but will have to rely on the mesh to get internet connectivity).
I think this is the way to go and see how things will work out.
You probably want to move this discussion to the ubiquiti forum.
This small forum here has been way more helpful than that one. Their forum is huge and hundreds of posts go unnoticed. Everyone's just very focussed on getting their issue sorted and general questions like this one are mostly ignored.
Also, I'm pretty sure I've figured out their controlling software and its capabilities after the few tests I've done so far.
If things go well, this project will start next February and I'll update this thread if thigns move forward.
Thanks for all the ideas and comments, you guys have all been very helpful.