Best architecture for a vision system

I want to create a vision mechanic, just like described in here — http://answers.unity3d.com/questions/41759/creating-enemy-field-of-vision.html — but in my game I have multiple enemies that can see things AND multiple friendly objects that can be seen by those enemies.

Should I use the method described in the link above, or just create field-of-vision meshes and use them as triggers? If I'll use the first one, what is the best way keeping organizing the data structure which would list all the possible friendly objects that all the enemies can look at?