The paper introduces the checkpoint notion for the management of service level agreement (SLA) information within dataflow property graphs. The purpose of checkpoints in an SLA graph database is to assist the automated handling of SLA resources with a flexible and expressive data structure, such as a multi-relational graph. We use an abstract scenario of cloud service provisioning for the analysis of the SLA and checkpoint graph data models. The paper integrates checkpoints into the graph schema by first establishing their relationships to internal SLA elements and to service management operations. In particular, the proposed graph data schema is used for the modeling and processing of a minimal service level evaluation schema. Checkpoints perform data value comparisons and return the service level evaluation status of SLA guarantees at defined time intervals. The SLA graph database is deployed as an RDF quad-store, where checkpoints represent service resources that can be evaluated dynamically and can be defined quite flexibly.