We address the problem of the planning and scheduling of technicians visits to customers on the field, in the context of service logistics over a multiperiod horizon. We have developed a VRP type memetic algorithm and a column generation/branch and bound heuristic in order to optimize the plan over a static horizon. We have then adapted our procedures to cope with a rolling horizon context, when a new plan has to be determined after the execution of each first daily period of the previous plan. We have tested our procedures on realistic data from the water distribution context, and obtained good solutions in a reasonable amount of time. We show in particular the advantage of reutilization of partial solutions from the previous plan, for the optimization of each new plan. Directions for further research are then indicated.