The azimuthal substructure observed in some debris disks, as exemplified by epsilon Eridani, is usually attributed to resonances with embedded planets. In a standard scenario, the Poynting-Robertson force, possibly enhanced by the stellar wind drag, is responsible for the delivery of dust from outer regions of the disk to locations of external mean-motion planetary resonances; the captured particles then create characteristic ``clumps. Alternatively, it has been suggested that the observed features in systems like epsilon Eri may stem from populations of planetesimals that have been captured in resonances with the planet, such as Plutinos and Trojans in the solar system. A large fraction of dust produced by these bodies would stay locked in the same resonance, creating the dusty clumps. To investigate both scenarios and their applicability limits for a wide range of stars, planets, disk densities, and planetesimal families we construct simple analytic models for both scenarios. In particular, we show that the first scenario works for disks with the pole-on optical depths below about ~10^{-4}-10^{-5}. Above this optical depth level, the first scenario will generate a narrow resonant ring with a hardly visible azimuthal structure, rather than clumps. The efficiency of the second scenario is proportional to the mass of the resonant planetesimal family, as example, a family with a total mass of ~0.01 to 0.1 Earth masses could be sufficient to account for the clumps of epsilon Eridani.