Atlas Performance Advisor helps identify slow queries and recommend indexes for the collections they affect. Its slow-query threshold adjusts to average operation times in the cluster, so recommendations reflect the workload. Each suggested index comes with sample queries grouped by query shape. section 21.2.1
Inspect recommendations with the Atlas CLI:
- Run `atlas processes list` to find a node name. The connection string identifies the cluster, but these commands target an individual process. section 21.2.1
- Pass that name as `--processName` to `atlas performanceAdvisor namespaces list` to inspect namespaces. section 21.2.1
- Use `atlas performanceAdvisor suggestedIndexes list --processName <node-name>` for index recommendations, or `atlas performanceAdvisor slowQueryLogs list --processName <node-name>` to inspect slow-query logs. section 21.2.1
Review an index recommendation against the queries it would help: extra indexes can slow writes and inserts. The Atlas-managed slow-operation threshold is enabled by default and can be disabled or re-enabled with `atlas performanceAdvisor slowOperationThreshold disable` or `enable`. If disabled, a fixed 100 ms slow-query threshold applies; Atlas-managed thresholds are unavailable on M0 and Flex clusters. section 21.2.1
Performance Advisor also offers schema suggestions for high-traffic collections and slow-running queries. In the Atlas UI, evaluate recommendations such as restructuring fields, adjusting data types, or improving indexes as workloads and document structures change. section 21.2.2
