The usage of cloud-computing architectures and characteristics has been enhanced in recent years. This approach brings the availability of storage and user services as needed. But it also brings many drawbacks that put the privacy and the security of the system at stake. Log file generation has Big Data characteristics that should be considered for upgrade from a manual method to an automatic one based on Big Data solutions. Therefore, this paper proposes a log file centralization and a diagnostic approach based on the misuse and the anomaly detection techniques, which will improve the detection of attacks. The FP-growth algorithm presented will ensure the prevention of consecutive violations.