Networked Control Systems (NCS) are commonly used in industrial applications like telerobotics, smart energy grids, and autonomous vehicles. In many cases, NCS share their network with other participants competing for the available bandwidth. This necessitates scheduling algorithms respecting the time-critical nature of control systems. Scientific evaluations under reproducible conditions are required to assess the performance of a given scheduling algorithm in the context of networked control systems. In this thesis, a framework to evaluate the performance of scheduling algorithms using a physical networked control system is designed and implemented. The framework comprises an inverted pendulum connected to an IEEE 802.3 Ethernet network featuring a software switch. The software switch can be programmed to execute arbitrary scheduling algorithms, significantly simplifying the evaluation process. This thesis explains the framework's design, implementation, and usage in detail. We use the proposed framework to evaluate the Multi-priority Token Bucket scheduling approach (MPTB) designed at the Institute for Parallel and Distributed Systems (IPVS) of University of Stuttgart. This scheduling algorithm dynamically assigns priorities to data streams according to their contract compliance. It is found that MPTB can provide better stability to the inverted pendulum at a lower average data rate than traditional FIFO scheduling. However, we also find that the selection of parameters for MPTB severely impacts the scheduling algorithm's performance. Further, we find that using real cross-traffic to stress the network yields non-deterministic latencies, while simulated delays at the software switch are better suited for reproducible evaluations.