Modern software systems continuously generate vast amounts of heterogeneous logs and measurement data, typically collected for monitoring, debugging, and diagnostics. These non-functioning logs often exist in disparate formats and lack temporal alignment, making them challenging to utilize directly for modeling and analysis. This work proposes a unified model that reshapes such raw software data into a time-series structure with minimal or no preprocessing required. By treating all logged information—regardless of source, format, or granularity—as temporal events, the model preserves the chronological context while normalizing irregularities in structure and frequency. The resulting time-series representation enables direct ingestion by time-aware machine learning models and facilitates downstream tasks such as anomaly detection, predictive maintenance, and system behavior modeling.