AI-Driven Chunked Data Migration for Active Read/Write Systems

The patent divides the data migration process into discrete event-driven transactions. Each data change event is captured, processed, and migrated independently as a separate transaction unit. This segmentation allows parallel processing of multiple events simultaneously, reducing overall migration time while maintaining data completeness through individual transaction integrity.

The system implements event-driven feedback mechanisms where data change events are continuously monitored and captured from the source system. Each event triggers a migration transaction that processes and transfers the specific data change. This feedback loop ensures that only committed, valid data changes are migrated, maintaining accuracy while allowing the source system to remain operational and accept new transactions throughout the migration process.

The patent transitions from static synchronous migration to dynamic asynchronous event-driven migration. The system dynamically processes data changes as they occur, adjusting migration timing based on event arrival and processing conditions. Multiple migration transactions can be in progress at different stages simultaneously, with the target system applying changes in the correct sequence based on event timestamps and dependencies, maintaining consistency without synchronous blocking.

The system extracts and processes only the specific data changes that have occurred since the last migration point, rather than migrating entire source files. Each event-driven transaction contains only the necessary data change information (insertions, updates, deletions), which is then applied to the target system. This extraction approach maintains data completeness while dramatically reducing the volume of data processed and the computing resources required.