To improve the quality of data-driven decision-making and insights, Airbnb implemented its Minerva system, which requires a series of data checks and anomaly detection for certain metrics. However, after several data engineering teams started building and implementing their own data check tools, the company decided to build a unified, extensible, config-driven framework for data validation – Wall. Wall, which is written in Python on top of Airflow but also supports SQL-based checks, decouples checks from DAG definitions, and ETL pipelines. Here, Subrata Biswas describes its architecture and key features.