Periodic broadcast is an effective paradigm for large-scale dissemination of popular videos. Considerable research efforts have gone into designing many excellent periodic broadcast protocols in terms of minimizing the server network bandwidth and the client resources. However, there are only a few implementations of periodic broadcast protocols available. This is probably because little has been documented on how the memory and disk bandwidth resources of a periodic broadcast server should be allocated. In this paper, we present a Generalized Periodic Broadcast Server (GPBS) model that supports any periodic broadcast protocol. Based on this model, we formulate and solve a new optimization problem whose solution provides insights into the server?s memory and disk resources allocation. We also discuss our prototype implementation of GPBS. Our work facilitates future implementation and deployment of many existing periodic broadcast protocols.