The company I work for (a WISP) intends to submit a tender to bid for a 'wireless network deployment' in a university campus with a land size of about 6 hectares.
The ffg are required for the deployment:

1. Blanket the whole campus environment with a metro wi-fi;

2. The deployment will support roaming for mobile surfing, VoIP for mobile phones, VPN for security, bandwidth intensive applications (video streaming, voice, and data).

We are looking at deploying wireless mesh routers at various points within the campus. In view of the above:

Q1. Going by the size of the campus (6 Hectares), like how many wireless mesh routers will be required to achieve this?

Q2. How far away from each other do we need to deploy these mesh routers considering overlap coverage area which will support roaming and handoff?

Hi,

Mikrotik is not good for school and company enviroment.
I guess you are gonna need some 802.1x authentication and stuff like that?

You can't do seamless roaming between AP's without centrally managed wireless switch and AP's as thin clients

Which country are you from? I can recommend some centrally controlled wireless AP's.

Also number of AP is very dependent on buildings on site, it could be 100AP or it could be easily 600Ap's

Are they going to be connected by ethernet?

Even if not, you can use a few WDS links to tie whatever isn't hardwired into a single network.

I disagree with the above poster. The only limitation I've found is what I can make the system do, not what it is capable of.

I've got a few hotels running MT, the least troublesome of all the WLAN's I've done.

Well hotel site is completely different network that school requires.

You had on that site one simple ssid with no authentication at all?

You simply cannot do 802.1x with seamless roaming and other stuff like that if you dont have centralized wireless controller.
I don't even mention automated channel and power asiigment that centralized wireless can do automaticaly?

Also personally if I could choose between managing 100 mikrotik AP's with same services or one wireless controller that will do all configuration and roaming for me guesss what I would choose ...

I should agree with you on the fact that for me to achieve seamless roaming (in a metro Wi-fi deployment scenario), I would need 802.1X authentication.

But I dont know if I should believe the opinion that I can't achieve seamless roaming without deploying a 'Centralised WLAN Controller and Thin APs'.

I tend to reason with the poster (Oldman) when he said ' The only limitation I've found is what I can make the system (MikroTik) do, but not what it is capable of doing'. He seems to make some sense.

How did you come to the conclusion that MikroTik cannot handle this kind of deployment?

I need further clarification on this ....Forum members, can I have your input?

from a management perspective, MikroTik is far from ideal. There is no centralised management for MikroTik (The Dude is mostly monitoring with some management bolted on) as there is for something like Cisco.

Having said that, I also don't agree that you should take a controller approach in the classical sense. Ye ol' controller setups insist on passing each packet through the controller's ports, which is very inefficient especially if you have a wide area. Go for distributed, intelligent APs but with centralised management. Ruckus has some good stuff.

MikroTik could fit in nicely at your core, handling all the RADIUS services, Hotspot login pages and all the fancy routing. You could also use MikroTik for the distribution network, MikroTik <> MikroTik PTP, PTMP and Mesh setups work very well and are cheap. But on the access layer (ie. the part feeding the end users), go for something better suited to that environment. Ruckus, Proxim, Cisco come to mind.

I completely agree with andreacoppini.

I did not even mention distributed (or local) forwarding functionality on AP's because I think all vendors implemented it.

Personally I have very good experience with Trapeeze wireless and H3C controllers.

Trapeeze has the best management and monitoring services, but it it's quite expensive.
H3C is little harder to configure and to monitor, but it's brand new product so it's only matter of time when they are gonna write these extra monitoring function's to the code

..............paraphrased:
The only limitation I've found is what I can make the system do, not what it is capable of........................

I've got a few hotels running MT, the least troublesome of all the WLAN's I've done.

Cant resist jumping in here...Every deployment should have it's 'setup' designed 'suitable for the purpose' of what exactly the customer wants. That makes OldMan's statement quite appropriate, and I'm sure everyone has one time or another been told what the 'best' system is. MT can do anything you know how to make it do. Sometimes thats good enough (suitable), sometimes its a nitemare(bad expectations), sometimes you dont know how it works so very well, and the customer loves it even more(more profit)...

MT is a powerful enabler of almost any network and/or wireless application. What is available from other sources is really just 'compiled code'. MT gives you assembler level capability, but you must 'make the system do' the job end-to-end by developing a customer and operator friendly environment. Some would just go get the shrink wrap, others would go find a project that uses what your 'system does'...Or you might not have the resources to make it do what it should. Maybe you need to do the project, or are a risk taker and like a challenge. MT can, will, and should be a component of 'the project' or possibly a very important point has been missed visiting this forum.

There are so many suitable opportunities that MT drops in, I would carefully review all projects for 'money pits' that the customer is specifying certain vendor specific functionality. Thats just a way for a bid to get the brand of equipment the customer wants, not what the bid requires...

Enough verbosity now....Had a very late network upgrade nite...Comments?...SZ

I find that MT is much more powerfull than available wiki/manuals would lead you to believe. That said MT comes with no guarantees as some niche specific hardware does therefore you ought to do inhouse testing before bidding or deployment. Than there's further testing of new firmware. With great power comes great responsibility.

There is a reason why other solutions are multiple of the cost of MT installation and not all of it is price gauging or marketing gimmics.