As the complexity of online services increases, there is a corresponding need for service-oriented systems to provide support for answering questions about how they have pro-
cessed and produced data. This need is particularly evident in compositions of services, where audits of each individual service’s use do not provide a connected picture of the com position’s processing history. The provenance awareness of a system is its ability to answer questions about the history of its processing, through recording provenance data during
execution. As the size and usage of a system increases, so can the size of the provenance data recorded, leading to increased demands on storage and decreased performance of
the service. However, the exact impact of provenance recording depends on what details of the service execution are being documented. Our goal is to make provenance awareness accessible as an explicit non-functional property (NFP) in composite service specications, as is common for performance or reliability properties. This would enable composite service designers to analyse the properties of their services to see whether they meet users’ requirements based on the provenance questions they could ask about the service’s
outputs and the dependent performance, storage and other properties. We present a preliminary approach towards this end, focusing on the step of categorising potential provenance questions according to their eect on other NFPs, so that the provenance awareness of a service can be specied as the categories of questions it can answer.