At its most basic, the motivation for continuity in real functions from R into R is "I can draw the entire graph of this function without lifting my pencil from the paper." Since we are not interested in the limitations/physics of pencils but in the behavior of the values in the range of the function when we vary variables in the domain of the function, we come up with a definition of continuity that basically says "when we move towards the point a0 in the domain of a function f, we also move towards the point f(a0) in the range of f." We then call f continuous at a0. If f is continuous on its entire range, we say f is continuous.
You can easily generalize this to multivariable and vector functions, where we basically are interested in functions without holes or jumps. Be careful with pathological behavior, however! there are simple functions that are only continuous at a single point, and there are functions that are mathematically continuous that do not look like anything you could ever draw with a pencil (ie., Weirstrauss).
Generally, we notice that functions behave "nicer" when they have more continuous derivatives. Ie., a functions which has continuous derivatives of the 5th order (any polynomial) is said to be smoother than a function that has discontinuous derivatives of the 3rd order (ie., f(x) = {x2, x >= 0; -x2, x < 0} ).
Functions that have continuous derivatives of all orders are generally said to be smooth, like polynomials. There are non-polynomial functions that behave almost as nicely as polynomials: it is well known that many functions are equivalent to their Taylor series on some neighborhood around the point that the series is based on. Functions that are equivalent to their Taylor series (kind of like an infinite polynomial) everywhere (technically, on some neighborhood of each point) are said to be analytic.
These definitions, used in real analysis, motivate the definitions in complex analysis.