Toward Widespread Remote Laboratories: Evaluating the Effectiveness of a Replication-Based Architecture for Real-World Multiinstitutional Usage

Educational remote laboratories, when properly designed, have been demonstrated to be highly effective from a pedagogical point of view. Throughout the years their technology has evolved and they offer great capabilities, that are sometimes even superior to hands on labs. Nonetheless, despite their potential, they are not yet a widespread tool across learning institutions. One of the main reasons is that many remote laboratories have traditionally been research-oriented and have not been able to guarantee a high-enough quality of service (QoS) in a real-world educational environment. Such a QoS requires a relatively high number of students being able to access the labs concurrently, since classes may include dozens of students. Likewise, such a QoS requires reliability: the laboratory must work, be available and provide correct results. This work evaluates whether by applying an architecture oriented towards cost-effective instance replication and a model oriented towards fault-detection, it is possible to create laboratories that can provide a high QoS and that can therefore be used in a real environment, across multiple institutions. The study encompasses a period of 736 days and over 72,377 laboratory sessions, and relies on real data, from multiple institutions, of professors and students using the LabsLand Intel DE1-SoC FPGA remote laboratory, of which many instances are deployed. The results show that the QoS did indeed meet very high standards, and that such an approach can indeed lead to trusted remote laboratories, appropriate for real-world educational usage, and their eventual widespread adoption.