dynamic report output caching

DQM in combination with dynamic cubes brings exceptional caching capabilities. One aspect of the report execution process which is not cached in this system is the rendering of report spec and XML. In order to determine if the query can consume a cached result set cognos renders the XML "string" to generate a hash key and checks it against previous hash keys - if it matches the cached result set is used. On complex queries the parsing of the report spec and rendering XML is a high CPU utilization process and on our system is taking approximately 4-5s at 30% CPU for a single report. Retrieval of the cached result set and rendering HTML is about 1s.

With an assumption that the underlying data or report object has not changed and security is valid, the combination of [report ID], [string representing a unique combination of selected prompt values], and [security] could effectively serve as alternate in determining query uniqueness (just as rendering the XML does). The advantage of using this method is that the XML doesn't need to be rendered from the report spec to retrieve any cached result sets and avoids this part of the report/query service.

In addition - instead of caching the result set, if the entire report output as an HTML object were cached the entire report and query service processes could be skipped and it would have the same effect as pulling a static HTML page.

Here are some of the components required
1. Mechanism to generate and store hash key from unique report-id and prompt value combinations and store
2. Mechanism to dynamically cache report output associated with the hash key
3. Mechanism to evaluate security objects for user running the report against user creating the cached object
4. Mechanism to enable report output caching
5. Mechanism(s) to clear report output caching (manual, event based, fixed intervals, on data source update, etc)
6. Mechanism to update any session specific data held within the report output.