DataHub

GraphQL Field-Level Performance Bottleneck

warning
latencyUpdated Feb 23, 2026

Slow GraphQL queries hiding which specific nested fields (lineage, schemaMetadata, owner) are causing latency, making optimization impossible without path-level instrumentation.

How to detect:

Monitor graphql_execution_time overall query latency showing degradation (p95 > 2s), but without path-level metrics revealing which resolvers are slow. Detect when overall query time exceeds thresholds but individual field-level timing is unavailable.

Recommended action:

Enable path-level GraphQL instrumentation with fieldLevelOperations targeting expensive operations (searchAcrossEntities, getDataset) and fieldLevelPaths targeting known expensive resolvers (/**/lineage/**, /**/relationships/**). Monitor graphql.request.duration with operation tags to identify which queries are slow, then use field-level timing to pinpoint exact resolver bottlenecks requiring caching or optimization.