Hi Dzianis,
Understood your strategy and agre with what you are doing.
1)
You can try the old move to SQL from SP7 with below hint. BTW: This is not the current SP8 behavior and probably will be disabled soon:
select * from _sys_bic."pack/view"
(placeholder."$$CE_INTERNAL_RELATIONAL_OPTIMIZATION$$" => '1')
1 is the instruction to convert to SQL (it can be a big big monster) like it was default on SP7 when flagged enforceSQL=true.
As this is just for internal "move to SQL engine" you will not see the SQL unless you provoke a runtime error and borrow the SQL from indextrace.
This case of one "source" serving these projections can provoke:
- loss of filter push downs
- materialization of MainView (if choose CE)
- high CPU (if choose SQL)
2)
The great news is that HANA DEV continue to improve performance on each version. I've notice another gain on rev82 and there are rumors that things will be even better on SP9 (time for enhancements).
What I particular like to do is investigate, isolate the undesired behavior and reproduce it in a simplified way and create a ticket to SAP so they have a chance to analyze and improve performance.
You can't count with such enhancements for your current project, but they will be analyzed and improved in some time. Just a sample some tickets I created around rev73 will be realeased for SP9.
Best regards, Fernando Da Rós