We don't know how the gravity really works. We don't know how anything really works. All we have are models that make predictions that we can test. As mathman pointed out, Newtonian Gravity gives you satisfactory approximations in most practical situations. Being far easier to compute, Newtonian Gravity is usually used. But General Relativity is perfectly adequate for describing all the same things. Yes, you can easily describe apple's fall from the tree with Einstein's equations as motion caused by curvature of spacetime. It's not even all that hard. But still, way more involved than same computation done according to Newton.
The reason we use Einstein's formulation to compute effects of gravity is because either Newtonian formalism doesn't provide sufficient precision (e.g. GPS navigation) or because it fails completely (e.g black holes). In other cases, we usually use Newtonian formalism because it's easier. But not because GR cannot be used. It always can be.
