Compare and contrast OLTP and OLAP systems in terms of their architecture and use cases.

Official answer

OLTP, or online transaction processing, emphasizes many short operational transactions: placing an order, updating inventory or recording a payment. Designs often favor predictable low latency, concurrent writes, enforceable constraints and selective indexed access.

OLAP, or online analytical processing, emphasizes broader reads, aggregations and comparisons over substantial history, such as quarterly sales by region. Designs often favor columnar scans, partition pruning, dimensional models and workload isolation. These are common patterns, not mandatory product categories: an analytical system can support transactions, and an operational system can run analytical queries.

A normalized operational schema can reduce update anomalies, while a curated fact/dimension model can simplify consistent reporting. Choosing either requires clear grain and definitions; denormalization does not automatically produce correct or fast analysis.

The systems often cooperate through batch or change-capture pipelines. Define the freshness target and reconcile late changes rather than assuming reporting is instantly current. Some platforms support mixed workloads, but capacity, contention and consistency still need measurement. Choose architecture from access patterns, correctness needs and cost, rather than calling every relational database OLTP or every NoSQL database OLAP.