Docs Menu
Docs Home
MongoDB Manual

Data Consistency

On this page

  • Use Cases
  • Tasks
  • Details
  • Data Staleness
  • Referential Integrity
  • Learn More

MongoDB gives you the flexibility to normalize or duplicate your data to optimize your application. If you duplicate data in your schema, you must decide how to keep duplicated data consistent across multiple collections. Some applications require duplicated data to be made consistent immediately, whereas other applications can tolerate reading stale data.

There are multiple ways to enforce data consistency in your application:

Performance Impact
Use Case
Updates to multiple collections occur in a single atomic operation.
Potentially high, due to read contention
Your application must always return up-to-date data, and can tolerate potential negative performance impact during periods of heavy reads.
Modify the application schema to embed related data in a single collection.
Low to moderate, depending on document size and indexes
Atlas Database Triggers
When an update occurs in one collection, triggers automatically update another collection.
Low to moderate, with potential delays in processing triggered events
Your application can tolerate reading slightly stale data. Users can potentially see outdated data if they run a query immediately after an update, but before the trigger finishes updating the second collection.

The best way to enforce data consistency depends on your application. To learn more about the benefits and implementation for each approach, refer to the corresponding documentation pages.

To enforce data consistency in your application, see the following pages:

The following factors can affect how you enforce data consistency.

Consider how important it is that your application returns the most up-to-date data. Some applications can return data that is minutes or hours stale with no impact to the user.

For example, in an e-commerce application, a user needs to know immediately whether or not an item is available. This information is ideally kept as consistent as possible, even if it requires frequent updates.

In contrast, analytic queries are typically expected to read slightly stale data. It is not critical to keep analytic data completely consistent.

Your application's tolerance for stale data affects how to best keep data consistent. Frequently updating data in multiple collections reduces the risk that a user reads stale data. However, frequent updates can negatively impact your application's performance. When you enforce data consistency, balance user needs with performance impact.

Referential integrity ensures that when an object is deleted, all references to that object are also deleted.

For example, an application has a products collection and a warehouse collection that contains references to the products collection. When a product is deleted from the products collection, the corresponding reference in the warehouse collection should also be deleted.

If your schema requires referential integrity, incorporate logic into your application to keep references consistent. At minimum, your application logic should prevent errors when attempting to query a reference that does not exist.


Handle Duplicate Data


Enforce Data Consistency with Transactions