It is very interesting question, I have not yet ever was able to hit the limit of users on ZD3000 system practically and it is quit difficult to check practically -- normally number of users even on big public events is less than you would expect - for 10 000 visitors event you rarely have more than 1000-2000 users simultaneously in fact.
Actually, I am not yet convinced by Michael answer (I have heard same statements before also) that this limit applies to all connected users.
This are my suggestions why I am not sure if it is that simple:
1. For me it looks more realistic that it is applied only to any kind of authenticated user, and, possibly to statistic subsystem, as users on open SSIDs (without encryption and authentication) are associated by AP without ZD intervention.
2. I suppose that there is a size limitation for user database on ZD, as well as some limits in databases used to maintain statistic data (such as a table of current users), but for ZD to count all active users and to force limit even on open SSIDs -- it would be a bit troublesome task to implement, and it provides no benefits (there is no additional license you can buy to add more users over 10000 limit anyway), so I don't see why Ruckus developers would waist time doing so.
3. It is also not clear, why this limit is needed at all? It is not performance thing, as ZD1200 has about 20x more performance than ZD1100, but it's stated limit is almost same. So, again, for me it looks that this limit is actually the size of user database in a software - but what if we use external RADIUS for authentication or no authentication as all?
4. There are some success stories about very high density cases published
by Ruckus on big stadiums, on older of them ZD3000 have been used, and I
think if this limitation would really exist, they would not work. So
probably it doesn't apply to unauthenticated users, at least not
enforced (it is possible, that statistics will not display user info
properly in such cases).
Anyway, all my suggestions about how it probably works are just speculations -- It would be really nice to get this question clarified once and for all by Ruckus developer team member -- who really knows firsthand how it is designed and how it works in reality. Can you get such answer for us from Ruckus developer team, Michael?
Also, it would be interesting to know if there is any such limitation
for vSCG? With 10000 AP limit per node, number of users supported must
be really huge, but SCG processing power per AP is much lower...
Of cause, in most cases this limit is irrelevant anyway -- you can hit this limit realistically only in very special projects, and it definitely doesn't apply to any office environment.
About providing Wi-Fi for events on temporary base, I completely agree with Primoz -- whenever you want to get high density, you want to use only directional models (ZF7762-S, ZF7782-S/N, T301S/N), and they work really great. You can easy accommodate 150-200 users on one AP, as far as you have correct design and tune configuration to disable slow connections and 802.11b devices.
If you are OK with R700, it means that you don't have really high density and than you don't need to care about user limit.
R700 is ok when you have big conference room with 1000 people but only 200 active Wi-Fi devices (actually 1x R700 can handle it reasonably). If you have 1000 mobile devices, you will be not very good with just 5x R700 (even so they will work much better than 5x APs from any competitor because of Beamflex+) -- but with 5x T301 you'll get much better total bandwidth and capacity.
If you have no idea about next venue, you probably want to use T301S mainly (30x120 degree diagram) and couple of T301N (30x30 degree). This makes very easy to cover any venue and you are much less constrained by AP location requirements. Or you can go with T301N only, which will provide the best possible performance, but than you'll need more of them to cover same area.