Direct Monitoring Dataset is a collection of data obtained during an experimental analysis of different direct monitoring techniques. The analysis based on a fault-injection campaign on a real-world air traffic control (ATC) middleware system conducted by SAFE tool. The transport layer of the considered middleware implements three direct monitoring techniques, i.e., event logs, assertions, and rule-based loging approach supported by LogBus. We injected 12,733 software faults, which types extend the Orthogonal Defect Classification (ODC) for practical injection purposes, into the middleware transport layer source code. For each injected faults, we collected the data generated by considered techniques at middleware level, but we also collected the event logs produced by the operating system (Red Hat 5 El) and the legacy ATC applications running on the top of the target middleware. In summary, collected data include: * data generated by middleware: * logs (MwLog) * assertions (MwAss); * rule-based logs (MwR-BLog); * operating system logs (OsLog); * application logs (AppLog). The dataset contains a folder for each technique mentioned above. Each folder contains 12,733 files, a file for each experiments (i.e., faults injected in the middleware). Each file cotains the data generated by technique when we inject that fault in the middleware transport layer source code. The name of file indicates the source code file where the fault is injected and the type of fault.