Network slicing is a fundamental capability for future 5G networks to properly support current and envisioned future application scenarios. Network slicing facilitates a cost-effective deployment and operation of multiple logical networks over a common physical network infrastructure such that each network is customized to best serve the needs of specific applications (e.g., mobile broadband, Internet of Things applications) and/or communications service providers (e.g., special purpose service providers for different sectors such as public safety, utilities, smart city, and automobiles). Slicing a RAN becomes particularly challenging due to the inherently shared nature of the radio channel and the potential influence that any transmitter may have on any receiver. In this respect, this article analyzes the RAN slicing problem in a multi-cell network in relation to the RRM functionalities that can be used as a support for splitting the radio resources among the RAN slices. Four different RAN slicing approaches are presented and compared from different perspectives, such as the granularity in the assignment of radio resources and the degrees of isolation and customization.