We have following questions regarding management of policy controlled sweeps:

Our starting position:

We have to implement up to 10 sweep policies in one objectStore working on document objects with different filter queries.

Findings:

The management of those policies is not transparent when the jobs are executed in scheduled time slots. We are not really aware of running status and finished status of each sweep policy. We are not able to receive the result set of the filter query in time even if we set a higher value for the property "filtered query timeout" (up to 240 sec). The policy controlled sweep examines all of the docversion entries and therefore needs a big amount of time to process a small amount of document candidates. Because of the underlying "OR" operator that is used for querying the Database when we have multiple sweep policies it is not trivial to set DB Indexes accordingly.

Our questions:

- In a future CPE Version, will there be a change in managing policy controlled sweep objects (i.e. more than one policy controlled sweep objects for document objects)

- Is this improvement already adressed in a planned release ?

- What is your best practice to achieve better manageable sweep policies now or in future releases ?

- What is your best practice to analyze the underlying queries which will be executed against CPE DataModel in order to set a matching DB Index / Query Plan ?

I feel with you and do not have a lot of answers.

In general with problems like this I had good experience in switching on DB Subsystem tracing and then running the operation in question.

Just be aware that the log is heavily convoluted ... but after a while you find the SQL in question.

Regards,

/Gerold