* Fix the application. This is usually what must be done. It's painful, requires meetings, can get very personal if the application developer is involved, and usually requires the application to be modified in some way to reduce the popularity of the single cloned buffer.

* Move the rows. If you're lucky, there may be multiple rows that are intensifying the buffer's popularity. If possible, distribute the rows, so multiple buffers are now not quite so popular. While altering the traditional pct_free and pct_used storage parameters is an option, for increased control, consider setting the maximum number of rows table blocks can hold. Unexpectedly, there is more involved than simply issuing an alter statement like alter table all_status minimize records_per_block 5. But it works wonderfully! For details, see a SQL reference guide or do an Internet search.

* Balance the workload. If you have control over the workload intensity, during peak cloning activity, consider reducing the workload specifically associated with buffer cloning. While not an exciting technical solution, workload balancing can have significant positive performance results.