Network slicing in 5G aims to provide an end-to-end partition of the physical network that is optimized for the service it has to supply. Each slice needs to fulfill a Service Level Agreement (SLA), that is a contract between the slice provider and the tenants on the quality of service and reliability, expressed for a diverse set of physical resources (spectrum, link capacity, computing power, etc). For the multi-resource allocation problem in network slicing, we provide two scheduling algorithms that take into account SLA requirements in terms of minimum and nominal resource quantity demands. We show that the algorithm that considers the availability rate of the service, in addition to providing the minimum capacity, has better performances in terms of time-fairness. For both scheduling algorithms we consider a user delaying policy able to take into account SLA priority and latency requirements.