For a buffer to be cloned, a process must be interested in an earlier version of the buffer. This implies two actions. First, the buffer must have been changed. Second, it is being queried. And for the contention to become significant, the buffer must be very popular; otherwise, only a few clones would be created. Popular buffers are popular for a reason. Usually, with the information shown in Figure 6-13, a DBA (or someone who knows the applications well) will know why the buffer is so popular. Because Oracle's hashing algorithm is based on the file number and block number, the solutions are frighteningly limited.

Figure 6-13. This OSM script is based on DBA_EXTENTS. For a given file number and block number, it produces information that can be used in a variety of situations, such as helping to resolve buffer cloning issues.

Please remember that having many clones may not present a performance issue. If it is in fact a performance problem, CBC latching contention will clearly be an issue. If this is the case and you spot a cloned buffer, then consider the following as possible remedies: