You have degraded, inconsistent reporting data from multiple on-premises sources. Following Google-recommended practices to detect and clean anomalies, what should you do?

The best practice for cleaning inconsistent data from multiple on-premises sources on Google Cloud is to first centralize it in Cloud Storage. Cloud Storage is a highly scalable and durable object storage service suitable for raw data ingestion. Once the data is in Cloud Storage, Cloud Dataprep is the recommended tool for visually exploring, cleaning, and preparing data for analysis. It's designed for data wrangling tasks, including anomaly detection and data transformation, without requiring extensive coding. Cloud Datalab (now superseded by AI Platform Notebooks) is primarily for data exploration, analysis, and machine learning model development using notebooks, not for large-scale data cleaning and preparation. Neither Cloud Datalab nor Cloud Dataprep directly connect to on-premises systems for data ingestion; data should be moved to Cloud Storage first.