Database administrators struggle when managing workloads that have widely different performance requirements. For example, the same database may support short-running OLTP queries and batch jobs containing multitudes of queries with varying complexity. Different workloads may have different performance requirements, expressed in terms of service level objectives (SLOs) that must be fulfilled in order to keep the issuing database users satisfied. In this paper, we identify basic query classes and describe the challenges they pose for SLO-aware workload management. Additionally, we propose a generic architecture for an SLO-aware DBMS. We give an overview of workload management techniques already implemented in today's DBMS and outline future research directions for as yet unsupported concepts.