# Atlas managed slow operation threshold and Performance Advisor index suggestions > Source-anchored research for Atlas managed slow operation threshold and Performance Advisor index suggestions. Run status: COMPLETED. Parent: [Database Profiler vs Query Profiler](https://llms-explorer.com/tree/database-profiler-vs-query-profiler/) · 4 facets · 20 facts · page: https://llms-explorer.com/tree/atlas-managed-slow-operation-threshold-and-performance-advisor-index-suggestions/ ## Measurements and reference values - MongoDB advises against fixed thresholds below 100 ms; a OneUptime tutorial instead illustrates 50 ms after disabling management, so its example should not be treated as MongoDB recommended practice. (medium confidence) (verified as of 2026-03-31) — [source](https://www.mongodb.com/docs/atlas/performance-advisor.md) ## Problems, failure modes and limitations - Atlas enables managed slow-operation thresholds by default on supported clusters and dynamically adjusts them using execution times across the cluster; 100 ms is the fixed fallback when management is disabled. (medium confidence) (verified as of 2026-10-14) — [source](https://www.mongodb.com/docs/atlas/performance-advisor.md) - A 2022 support response says query frequency affects suggestions, so infrequent slow queries need not produce recommendations. (low confidence) — [source](https://www.mongodb.com/community/forums/t/alert-all-the-time-query-targeting-scanned-objects-returned-has-gone-above-1000/168277) - MongoDB Atlas Query Insights templates read Performance Advisor slow logs and suggestions to report on query shapes; the template cannot create indexes. (low confidence) (verified as of 2026-10-30) — [source](https://github.com/mongodb/agents) - Inference for safe index removal: absence of an Advisor suggestion cannot prove safety because recommendations depend on query frequency and captured slow-log evidence. (medium confidence) — [source](https://www.mongodb.com/docs/atlas/performance-advisor.md) - The project-level Managed Slow Operations toggle and the CLI commands atlas performanceAdvisor slowOperationThreshold disable and enable control automatic threshold management. (medium confidence) (verified as of 2026-10-14) — [source](https://www.mongodb.com/docs/atlas/performance-advisor.md) - Performance Advisor suggests indexes for slow queries and includes sample queries grouped by query shape. (medium confidence) (verified as of 2026-10-14) — [source](https://www.mongodb.com/docs/atlas/performance-advisor.md) - Atlas does not support the managed slow-operation threshold on clusters with MongoDB Search enabled, and Free clusters cannot enable it. (low confidence) (verified as of 2026-10-14) — [source](https://www.mongodb.com/docs/atlas/performance-advisor.md) - Inference: absence of an index suggestion cannot establish workload-wide safety for an index change, because recommendations depend on observed slow operations and log coverage. (medium confidence) (verified as of 2026-10-14) — [source](https://www.mongodb.com/docs/atlas/performance-advisor.md) ## Comparisons and alternatives - Performance Advisor suggests indexes from observed slow queries; suggestions reflect captured workload rather than every possible query. (medium confidence) (verified as of 2026-10-30) — [source](https://www.mongodb.com/docs/atlas/performance-advisor.md) ## Facts and statements - Review write costs before adding suggested indexes; maintaining more indexes can reduce write performance. (medium confidence) — [source](https://www.mongodb.com/docs/atlas/performance-advisor.md) - Suggestions include sample query shapes and impact ranking based on wasted bytes read. (medium confidence) (verified as of 2026-10-30) — [source](https://www.mongodb.com/docs/atlas/performance-advisor.md) - Extreme log volume can interrupt Advisor log collection; downloadable logs remain complete. (low confidence) (verified as of 2026-10-30) — [source](https://www.mongodb.com/docs/atlas/performance-advisor.md) - Managed thresholds are unsupported with MongoDB Search enabled; Performance Advisor requires M10+. (low confidence) (verified as of 2026-10-30) — [source](https://www.mongodb.com/docs/atlas/performance-advisor.md) - In a worked Java application example, requests filtering movies by title and year produced an Advisor suggestion for {title: 1, year: 1}. This illustrates workload-driven suggestions, not a general benchmark. (low confidence) — [source](https://nljug.org/foojay/why-mirroring-production-in-dev-helps-you-avoid-costly-mistakes/) - Performance Advisor is available on M10+ clusters; recommendations consider replica-set workload activity including secondaries. (low confidence) (verified as of 2026-10-14) — [source](https://www.mongodb.com/docs/atlas/performance-advisor.md) - Suggested-index metrics include execution frequency, average execution time and documents examined per document returned; Impact ranks suggestions by wasted bytes read. (low confidence) (verified as of 2026-10-14) — [source](https://www.mongodb.com/docs/atlas/performance-advisor.md) - Before creating another index, weigh its write cost and query frequency, and consider whether existing indexes can serve a modified query. (medium confidence) (verified as of 2026-10-14) — [source](https://www.mongodb.com/docs/atlas/performance-advisor.md) - Index-suggestion coverage has limits: time-series collections receive no performance suggestions, and extreme log volume can interrupt Atlas log collection for these interfaces; downloadable logs remain complete. (low confidence) (verified as of 2026-10-14) — [source](https://www.mongodb.com/docs/atlas/performance-advisor.md) - For its own log-based compatibility analysis, AWS documents disabling Atlas threshold management, enabling full logging with db.setProfilingLevel(0, -1), then restoring managed operation afterward; it warns of overhead and recommends dev/test capture. (low confidence) (verified as of 2026-10-30) — [source](https://github.com/awslabs/amazon-documentdb-tools/blob/master/compat-tool/README.md)