Supply chain optimization is a very complex task when taking into account its inherently decentralized aspect and sustainable development considerations. In order to provide a sustainable supply chain model faithful to reality, this paper presents a two actors supply chain model with a manufacturer and a 3PL provider. Centralized and decentralized planning at tactical level, with lot-sizing models, presented by Jung et al. (2008) is studied. This model is extended by including rolling planning horizon, and demand forecast errors, to take into account difficulties of supply chain actors to evaluate future demands. A preliminary experimental study highlights the good economic performance of the decentralized planning and the consequences of demand uncertainty on supply chain total cost. Then, we point out the poor production resource management with rolling horizon planning. Because of social consequences that may result, a new constraint is integrated in the supply chain model: securing production resources. This results in the impossibility to modify the planning on the coming periods. The economic impact of the integration of this social criteria is studied. Results are presented based on deterministic or stochastic demands, and with different rolling horizon length. It points out that securing production resource is not very expensive when planning on a small rolling horizon or with deterministic demands. Various future extensions of this model are also discussed.