For example say you wanted to draw a 3D cube you can define an array of hard coded vertices using float3, here is an example. http://msdn.microsoft.com/en-us/library/windows/desktop/jj552949.aspx
But you obviously can't define by hand large and complex changing geometry vertices such as you would have in a game. So I am trying to understand the mechanics behind how these vertices are loaded/interpreted into the Engine.
Another example, you could create a model in say Maya and export this vertex data to say an XML file, then load up the model using GL/DX. So Maya is doing all the work determining the vertex data. You could do this for terrain also, any vertex geometry.
However a game state is constantly updating, it is not static, the camera is constantly moving with user input. So what is the mechanics behind constantly re-evaluating new vertex data depending on user input.