The management of voluminous, heterogeneous, and distributed data on a large-scale raises new problems and presents real challenges: efticiency of access, placement and redistribution of data, communication, confidentiality of access, availability of data, indexation, caching and replication mechanism, and benchmarking. In this paper, we point out the contributions of execution models based on mobile agents and cost model approaches to the large scale distributed query optimization. Firstly, we summarize state of the art of the main execution models based on mobile agents. We show how mobile agents can react dynamically to estimation errors, and resources unavailability. Then, to guarantee the autonomous of mobile agents, a part of cost model must be embedded. Finally, we study, the efticiency of mobile agents in relation with the cost model approaches, the query type, and the adequacy of the cost model with regard to the data sources. Identifcation of intervals for mobility eflciency, allows to define and validate a proactive migration policy.