The archetypical debris disk around Vega has been observed intensively over the past 25 years. It has been argued that the resulting photometric data and images may be in contradiction with a standard, steady-state collisional scenario of the disk evolution. In particular, the emission in the mid-infrared (mid-IR) appears to be in excess of what is expected from a "Kuiper belt" at ∼100 AU, which is evident in the submillimeter images and inferred from the majority of photometric points. Here we re-address the question of whether or not the Vega disk observations are compatible with a continuous dust production through a collisional cascade. Instead of seeking a size and spatial distribution of dust that provide the best fit to observations, our approach involves physical modeling of the debris disk "from the sources." We assume that dust is maintained by a belt of parent planetesimals, and employ our collisional and radiative transfer codes to consistently model the size and radial distribution of the disk material and then thermal emission of dust. In doing so, we vary a broad set of parameters, including the stellar properties, the exact location, extension, and dynamical excitation of the planetesimal belt, chemical composition of solids, and the collisional prescription. We are able to reproduce the spectral energy distribution in the entire wavelength range from the near-IR to millimeter, as well as the mid-IR and submillimeter radial brightness profiles of the Vega disk. Thus, our results suggest that the Vega disk observations are not in contradiction with a steady-state collisional dust production, and we put important constraints on the disk parameters and physical processes that sustain it. The total disk mass in 100 km-sized bodies is estimated to be ∼10 Earth masses. Provided that collisional cascade has been operating over much of the Vega age of ∼350 Myr, the disk must have lost a few Earth masses of solids during that time. We also demonstrate that using an intermediate luminosity of the star between the pole and the equator, as derived from its fast rotation, is required to reproduce the debris disk observations. Finally, we show that including cratering collisions into the model is mandatory.