Scenario:
Large system (~ 200 tables).
60,000 users.
Complex reports that will require me to execute several queries for each report, and even these will be complex queries with internal queries everywhere + some PHP processing.
An approach:
I saw an approach that I'm not sure about:
Having one centralized, non-standardized table that registers any activity in the system that is subject to reporting. This table will mainly contain foreign keys, so it should be fairly compact and fast.
So, for example, (My system is a virtual learning management system), the User registers in the course, the user identifier, date, course identifier, organization identifier, type of activity (registration) are stored in the table.
Of course, I also save this data in a normalized database that the actual application uses.
Pros: simple, supported queries and code for data processing and quick retrieval.
Cons: there is a danger that the de-normalized table will not be synchronized with the real database.
Is this approach being considered, or (preferably from experience) the sum of $ #% #% t?
source
share