Service-oriented architecture (SOA) is a design approach for the development, integration, and maintenance of complex enterprise information systems composed of loosely-coupled services. Runtime services composition introduces performance overheads. Software developers may have difficulties in assessing the performance costs of the final composite product. One objective of this thesis is to design and implement a unified logging facility, which allows us to instrument different layers of a SOA system under study, to automatically collect measurement data and to correlate the separate logs from sub-systems and components into a unified system view. Next, the logging facility is used to collect performance measurement data for a case study system. The measurements are collected separately for the application code and for the underlying service engine. The measured performance data is used for building a Layered Queuing Network model of the case study system and to validate it against measurements.