Using reports as source for a warehouse or other reports is a very poor design pattern, but currently its our only option. My question is, generally speaking, is it better to put a halt to a project if the data cant be directly accessed, or only accessed via a reporting system?
Detail: There are a number of pillar systems in my organization that contain important operational data on clients, orders, programmes etc etc
The systems are cloud based with either limited or no direct access to data, i.e. API's are incomplete & there is no direct access allowed such as ODCB.
The systems have their own reporting tools, Oracle Discoverer, Jaspersoft & proprietary tools. The suggestion is to extract reports from these systems on a regular basis and import the data from these sources to a data warehouse/mart.
As reports are generally much more fluid and rarely have an agreed format/schema, i foresee a lot of support calls and issues with any system that uses reports as its data source.
Tools like PowerBI can use reports and other oddly shaped data relatively easily, but i never saw it as the basis for an organizational data warehouse. Because of this however, I dont want to rule it out completely just yet, and look like a blocker in the organization.
There might be a middle ground.
Reporting tools like Oracle Discoverer and Jaspersoft can usually export their data in formats such as CSV or Excel.
You could use this feature like a simple ETL tool, writing custom "reports" that are really just extracts of normalised data to load into your warehouse, Azure Analysis Services, or PowerBI.