In Business Intelligence (BI) architectures, do BI systems avoid sourcing data from operational databases, or do they typically extract from OLTP systems via ETL/ELT?

Introduction / Context:
BI solutions rely on accurate, timely business data. Typically, that data originates in operational (OLTP) systems such as ERP, CRM, HR, and order management. The BI stack extracts these records and transforms them for analytics, proving that BI indeed obtains data from operational databases.

Given Data / Assumptions:

• Operational systems store transactions: orders, invoices, shipments, tickets.
• BI requires curated, integrated, historical views.
• Standard pipelines: ETL (Extract-Transform-Load) or ELT (Extract-Load-Transform).

Concept / Approach:
BI ingestion connects to operational sources, captures deltas (change data capture, timestamps, or log-based replication), and conforms dimensions (e.g., Customer, Product) into a warehouse or data lakehouse. Reporting and analytics then consume these curated layers, not the volatile OLTP tables directly.

Step-by-Step Solution:
Identify operational sources and key entities.Extract data incrementally to staging.Transform and conform dimensions and facts.Load into warehouse/mart optimized for analytics.Serve reports, dashboards, and models from the curated store.

Verification / Alternative check:
Examine lineage in a typical BI tool: dataflow diagrams show operational systems feeding a warehouse/mart before reports.

Why Other Options Are Wrong:
Saying BI never uses operational sources contradicts core BI practice (option b). Cloud vs. on-prem does not change the fundamental flow (option c). Read-only status is not a requirement (option d). Normalization level in sources is irrelevant to the fact that BI sources them (option e).

Common Pitfalls:
Pointing reports directly at OLTP and causing contention; skipping data quality steps; misunderstanding that BI “creates” data rather than integrating it from operations.

Final Answer:
Does not apply — BI systems do source from operational databases