The AWR report indicates that the main Data Pump worker process, represented by the SYS.KUPW$WORKER.MAIN procedure, is consuming significant time. However, this is a symptom of deeper issues rather than the cause itself. The true bottlenecks are revealed in the top wait events, which show that the process is heavily constrained by high CPU utilization (49% of DB time) and significant I/O waits for db file sequential read (41.2% of DB time). This combination suggests the import is performing computationally intensive work, such as creating complex indexes or processing triggers, while also being limited by the speed of single-block read operations from the database datafiles on disk. To resolve this, I recommend a multi-faceted approach: first, increase the job's parallelism by setting the PARALLEL parameter in your impdp command to a value appropriate for the server's CPU core count (e.g., PARALLEL=8) to distribute the workload. Second, defer non-essential operations by using the EXCLUDE=INDEX,STATISTICS parameter to load the table data first, then creating indexes and gathering statistics in a separate, optimized step. Finally, for environments where recoverability can be managed by re-running the import, consider reducing logging overhead with the TRANSFORM=DISABLE_ARCHIVE_LOGGING:Y parameter to lessen the I/O burden. Implementing these changes will directly address the CPU and I/O constraints and should significantly improve the import performance.

Cheers

Arsalan