Quick answer
MongoDB Atlas documents read and write operations as 100 operations per second. Atlas throttles and applies a one-second cooldown when the limit is exceeded.
Verified Aug 22, 2026Official source
Current limits
| Constraint | Current value | Scope | Verified source |
|---|---|---|---|
| read and write operationsAtlas throttles and applies a one-second cooldown when the limit is exceeded. | 100 operations per second | Free cluster | MongoDBAug 22, 2026 |
Why does this limit matter?
Queued operations can take longer than one second even after a brief traffic burst.
This value is scoped to Free cluster; a different plan, runtime, model, endpoint, region, or account can produce a different effective constraint.
What should you check?
- Measure operations per second, queueing latency, and working-set behavior.
- Confirm the exact plan, model, runtime, endpoint, region, and account that serve the failing workload.
- Record the observed value, response headers or configuration, timestamp, and source without logging secrets.
Important caveats
- The operation rate is separate from data transfer and query-utilization limits.
- Treat the official source and live account configuration as authoritative if they differ from this verified snapshot.
HyperObserve reports the documented platform constraint. Your application, SDK, gateway, provider, region, or account can impose a lower effective limit.
Related