The library / Analytics & warehouses

OLTP and OLAP: two workloads, different priorities

Understand why checking out a shopping basket and analyzing a year of sales ask very different things of a database.

Start with the work

An application might create an order, reserve inventory, and read one customer's profile. A report might scan millions of orders to calculate monthly revenue.

Both use data, but their access patterns and operating priorities differ.

The transactional side

Online transaction processing, or OLTP, commonly involves many small reads and writes, concurrent users, and correctness requirements around individual business operations.

A useful design question is: how does the system keep one order correct while other orders are changing?

The analytical side

Online analytical processing, or OLAP, commonly involves scanning, joining, and aggregating substantial portions of a dataset.

A useful design question is: how can the system calculate a large result efficiently without interfering with the application's operational workload?

Question Transactional workload Analytical workload
Typical scope A small set of records Many records and aggregates
Example Update one order Revenue by month and region
Common priority Predictable transaction latency Efficient scans and computation
Design focus Constraints and access paths Data layout and aggregation

These are common patterns, not absolute boundaries. Some engines support mixed workloads, and small analytical queries may run well in an application database.

Decide when to separate

Separation introduces ingestion, duplicate storage, freshness lag, and another system to operate. It becomes useful when analytical work disrupts transactional performance or when analytics needs a different data organization.

Before moving a report, measure its actual impact and define how fresh the analytical result must be.

Follow the data

A typical flow starts with application transactions, then captures or exports changes, transforms them into analytical models, and serves reports.

Each boundary needs an answer for schema changes, late data, corrections, and reconciliation. Choosing a warehouse does not solve those pipeline questions by itself.

Keep exploring

Go deeper with the original documentation.

Official documentation
D
DBMinutes Editorial

Practical explanations of database systems, cloud services, and the engineering decisions between them.

AI-assisted content · Our editorial process

Keep the curiosity going.

Back to the library
A little learning goes a long way

Make room for a few good minutes.

Join the list for practical guides, thoughtful comparisons,
and ideas worth bringing to your next project.

Find your next answer

Search concepts, tools, and practical guides.