Problem
When analyzing distributed traces in Dynatrace, the waterfall view can become difficult to read because it displays all spans captured within a trace, including infrastructure and client transport calls.
For example, traces often contain:
- HTTP client calls to metadata services
- Service-to-service transport spans
- SDK/framework-generated client spans
- AI platform/API client calls such as *.aiplatform.googleapis.com
These spans are valuable for troubleshooting and should remain stored and searchable. However, during day-to-day root cause analysis they frequently add visual noise and make it harder to focus on the business transaction flow.
Requested Enhancement
Add filtering capabilities directly within the Distributed Tracing waterfall view, allowing users to:
Filter spans by
- Span kind (Client, Server, Internal, Producer, Consumer)
- Span name
- Service name
- URL or endpoint pattern
- Custom span attributes/tags
Hide spans without deleting data
Users should be able to temporarily hide matching spans from the waterfall visualization while preserving the underlying trace data.
Examples:
- Hide all Client spans
- Hide spans where URL contains ###.###.#######
- Hide spans where URL contains ###.###.###.###
- Hide transport or framework-generated spans
- Create custom include/exclude rules
Save Filter Preferences
Allow users to:
- Save personal filter configurations
- Reuse filters across trace investigations
- Set a default waterfall filter that is automatically applied when opening traces
Expected Benefits
- Improved trace readability
- Faster root cause analysis
- Reduced visual clutter in complex distributed traces
- Better user experience without sacrificing observability data
- No need to rely on external queries or separate span views for common filtering use cases
Important Note
This request is not about dropping, sampling, or removing spans from ingestion. The goal is only to provide a visualization-layer filter in the waterfall view while retaining complete trace data for later investigation.
