Database caching is operational, but nothing to write home about. For some basic page rendering queries in PBooks, the caching saves about .02 seconds. The value of this is questionable, considering the time required to render XSL pages, which can also be cached. In certain circumstances, there will be times when database caches will be useful - remote databases, similar datasets with different XSL renderings. For now, I'm putting database caching on the shelf. The process was helpful to clarify how the MDB2 API works. Well I did a little bit more work, it actually is helpful to some degree, and is easy to store and erase the caches, so its one more option to use. The flip side to the fact that the dataset cache doesn't have much of a speed advantage is that it is incredibly granular. The cache id's are set by the query name and the variables used in the prepared query, so its as detailed as the SQL itself, without being so bulky.