Networks in industrial control and automation applications must be capable of providing highly predictable delivery, both in terms of latency and latency variation. Thus, a network must guarantee deterministic behavior for time-sensitive traffic. Currently, dedicated and highly engineered networking solutions such as field buses are commonly used for time-critical traffic in industrial applications. Current trends in network convergence combine mixed-criticality traffic flows into the same real-time-capable networks. Currently, the IEEE Time-Sensitive Networking (TSN) Task Group are augmenting the widely supported Ethernet standard to provide time-critical flows alongside best-effort traffic. For its capabilities in high bandwidth, mixed-application and real-timesupport, TSN is expected to be widely adapted in industries. In TSN, many of the guarantees on deterministic behavior have their basis in a traffic scheduling mechanism called Time-Aware Shaper (TAS). The development and integration of TSN-capable networks require increasingly powerful diagnosis and measurement tools. However, no such framework is publicly available at this point. The focus of this thesis is the development of a framework fulfilling requirements for performance analysis of physical networks with real-time constraints. This framework enables high-performance measurement tasks for TSN, which makes measurable the inaccuracies of physical network devices. As a first result of this insight, we show that many of the assumptions researchers often make about TSN networks are simplified and do not properly model the real world. Moreover, scheduling algorithms for the TAS often disregard the same inaccuracies in the synthesis of the traffic schedule. This could render invalid the guarantees on determinism for out-of-the-box scheduling. We propose a mechanism to improve the determinism of low-quality senders and the overall network bandwidth utilization, proving its effectiveness in a proof-of-concept setup.