Patent attributes
Techniques are described for analyzing usage of data stored in a data storage system without accessing the stored data. In some embodiments, workload data indicative of queries executed at the data storage system on stored data is received. This workload data can include query logs generated during execution of the queries. The workload data is processed to identify data elements such as tables, columns, and views associated with the stored data as well as information regarding usage of the identified data elements. Usage can include operations performed on the data elements during execution of the queries. Based on this processing relationships between the identified data elements can be inferred and visualizations generated that convey information regarding usage of the data stored at the data storage system. Visualizations can include, among others, usage heatmap diagrams, join diagrams, column family diagrams, filter diagrams, view lineage diagrams, data flow diagrams, denormalization diagrams, and workload distribution diagrams.
