There are four straightforward steps to ensure the objects you want are always kept in the shared pool. While the word pin is commonly used, the dbms_shared_pool package's keep function is used to ensure the object is kept in the shared pool. This package is not loaded by default when the database is created, so your first step is to load it. The following code snippet is an example of how to create the procedure.

The next step is to find the large and/or popular objects. Oracle keeps track of shared pool object usage and makes this information easily available via the v$db_object_cache view. Figure 7-14 shows the output from the OSM script dboc.sql, which is used to identify potential objects. You're likely to see a group of packages that are much larger than the rest and also packages that are executed much more frequently than others. There may also be objects that you personally know have unusual execution profiles and you want to have cached.

Once you have the list of objects to keep, the next step is to determine how to place them in the cache. The keep function is used to pin the object, or better said, to keep the object in the shared pool.