Theoretical approach to Improve the Performance of APDs
Pre requisites: The reader understands and has extensively used APDs. 🙂 You just need the basic knowledge of an APD.
1. APD’s are not designed to work with large amounts of data. Make sure you are filtering out all the data that is not needed, let us say you need to quickly dig into past 2 months of invoices data-> do not pass all your historical invoices data through your APD.
2. Remove the option in the performance settings- > process in memory, if the data amount is still large enough -> 2 months of invoices data is big 🙂
3. Try to schedule the APD as a background job through the program- > RSAN_PROCESS_EXECUTE.
4. If you are using the infocubes or multicubes as your apd source, try partitioning the Infocube and compressing the requests, if the apd has to mine 2 months of data from f- table it has to query on all the requests, instead of the partitions (calmonth or fiscper)
5. table partitioning and compression, is a recommended practice, as a part of administration, which would also improve your apd performance.
Next Blog: To be determined..