20240045858. UPDATE PROPAGATION IN A DATA STREAM WAREHOUSE simplified abstract (AT&T Intellectual Property I, L.P.)

UPDATE PROPAGATION IN A DATA STREAM WAREHOUSE

Organization Name

AT&T Intellectual Property I, L.P.

Inventor(s)

Theodore Johnson of New York NY (US)

Vladislav Shkapenyuk of New York NY (US)

Divesh Srivastava of Summit NJ (US)

UPDATE PROPAGATION IN A DATA STREAM WAREHOUSE - A simplified explanation of the abstract

This abstract first appeared for US patent application 20240045858 titled 'UPDATE PROPAGATION IN A DATA STREAM WAREHOUSE

Simplified Explanation

The abstract of the patent application describes architectures and techniques for efficiently updating derived data products in response to updated source data. It explains that source data is typically stored in source tables, while a materialized view of a query can generate a derived table based on the state of the source tables at the time of execution. When the source data changes, an invertible relationship between timestamps can be used to identify only the affected portions of the derived table, allowing for the generation of a new defining query to update only those portions.

• The patent/application presents architectures and techniques for updating derived data products efficiently.
• Source data is stored in source tables, and a materialized view of a query can generate a derived table.
• An invertible relationship between timestamps is leveraged to identify the affected portions of the derived table.
• A new defining query can be generated to update only the affected portions of the derived table.

Potential Applications:

• Data analytics and reporting systems
• Real-time data processing systems
• Business intelligence platforms

Problems Solved:

• Reducing the computational cost of updating derived data products
• Minimizing the need for re-computation of the entire derived table
• Efficiently handling changes in source data

Benefits:

• Improved performance and efficiency in updating derived data products
• Reduced computational resources required for updates
• Real-time or near-real-time updates of derived data products

Original Abstract Submitted

architectures and techniques are presented that can more efficiently update derived data products in response to updated source data. source data is typically stored in source tables, whereas a materialized view of a query can generate a derived table based on the state of the source tables at the time the query is executed. when source data changes (e.g., in response to late-arriving input data), rather than recomputing the entire derived table (e.g., by again executing the original query, which can be expensive), an invertible relationship between timestamps can be leveraged to identify only those portions of the derived table that are affected by the update. therefore, a new defining query can be generated to update only those portions of the derived table that are affected by the source data update.