This paper reviews the literature on inventory models with multiple sourcing options and presents a typology for classification. By means of the classification, the progression of the literature (policies and modeling assumptions) is illustrated, the main decision trade-offs in multiple sourcing are identified and avenues for future research are pointed out. Multiple sourcing decision models trade off the added costs of backup sourcing against higher inventory or shortage costs under single sourcing. The value of multiple over single sourcing is found to increase in the uncertainty to be buffered, in inventory holding and shortage costs, as well as in the constraints of the primary source. The literature evolved from small, restrictive models to larger problems and more realism. Accordingly, replenishment policies progressed from optimal policies to more heuristic decision rules. Three areas for future research are suggested for moving the field forward and towards more practical applicability. (1) Further integration of model aspects such as the extension of replenishment policies to more than two suppliers and to multi-echelon models. (2) Focusing on supply chain resilience with decision making disruption events or demand spikes under consideration of risk preferences. (3) Utilizing industry data in machine learning and data-driven methodologies.

本文回顾了有关具有多种采购选项的库存模型的文献，并提出了分类的类型学。通过分类，说明了文献（政策和建模假设）的发展，确定了多源采购中的主要决策权衡，并指出了未来研究的途径。多种采购决策模型权衡了备用采购的额外成本与单一采购下较高的库存或短缺成本之间的关系。发现单一采购的多重价值增加了​​要缓冲的不确定性，库存持有和短缺成本以及主要来源的约束。文献从小的限制性模型发展为更大的问题和更多的现实主义。因此，补货政策已从最佳政策发展为更具启发性的决策规则。提出了三个未来研究领域，以推动该领域的发展并朝着更实际的适用性发展。（1）进一步整合模型方面，例如将补货政策扩展到两个以上的供应商和多级模型。（2）在考虑风险偏好的情况下，着重于决策中断事件或需求激增的供应链弹性。（3）在机器学习和数据驱动的方法中利用行业数据。（1）进一步整合模型方面，例如将补货政策扩展到两个以上的供应商和多级模型。（2）在考虑风险偏好的情况下，着重于决策中断事件或需求激增的供应链弹性。（3）在机器学习和数据驱动的方法中利用行业数据。（1）进一步整合模型方面，例如将补货政策扩展到两个以上的供应商和多级模型。（2）在考虑风险偏好的情况下，着重于决策中断事件或需求激增的供应链弹性。（3）在机器学习和数据驱动的方法中利用行业数据。