The Trace Explorer in Observe by Snowflake enables you to investigate user requests and transactions by analyzing trace data. With powerful filtering and visualization tools including the flame chart and waterfall view, you can connect these requests to the underlying services, infrastructure, or other resources that processed them. Whether you’re troubleshooting errors, inspecting performance bottlenecks, or analyzing specific requests, the Trace Explorer helps you quickly pinpoint and resolve issues.

For more information on how Observe by Snowflake makes trace data available, see APM reference.

Access the Trace Explorer

To access the Trace Explorer, click Traces in the left navigation rail in Observe by Snowflake.

Filter what you see in the Trace Explorer

Use the query panel to filter what you see in the Trace Explorer. For example, we can view only frontend services. When you select a filter in the query panel, the query is automatically updated:

You can also adjust the time window for your query using the time-picker dropdown. Select any of the pre-populated options, or click Calendar to customize a specific time window.

When you are done building your query, you can click toggle query panel icon () to dismiss the query panel.

Export data

To download the data displayed in Trace Explorer, click Export. You may select CSV or JSON format, and a maximum size limit (one thousand, ten thousand, or one hundred thousand rows). Note that hidden fields will be included. Use the pick_col OPAL verb to reduce the width of downloaded data.

Link directly to a trace

You may want to include direct links to specific traces from your alert payloads. To create a URL to a trace, use the following template:

https://<tenant_id>.observeinc.com/workspace/<workspace_id>/trace-inspector?traceId=<trace id>

Traces accessed via direct link are fetched from the last 24 hours of data.

If you are using a Dataset other than the Span Dataset that is part of the OpenTelemetry app, you'll need to add the Dataset ID as a parameter:

https://<tenant_id>.observeinc.com/workspace/<workspace_id>/trace-inspector?spanDatasetId=<span dataset id>&traceId=<trace id>

Examine trace details

Click View for any trace in the Trace Explorer to view its details. By default, a waterfall visualization of the trace is presented:

Select any of the tabs to change the visualization:

Click Add columns to add additional span attributes that are not in the default waterfall view, so that you can correlate attributes across spans without having to open multiple individual spans.

For example, you can add the service.namespace column to understand flows relative to broader business domains and organizations, or add server.address to identify when a call is going to an external service, like a third-party API.

After you add additional columns, you can slide the attribute's toggle on or off to control its visibility in the chart, or delete the attribute so that it's no longer part of the chart:

Click Ask o11y to get an AI-generated summary of the trace, including some recommended actions:

Use the Trace Map to view your services and calls

The trace map shows every service in a trace and the calls between them. That includes inferred services such as databases, message brokers, and uninstrumented dependencies. Use it to understand how a request flows through your system and to decide where to start investigating complex distributed traces.

What you can do:

• Find hotspots: See where errors are concentrated and where most of the trace's duration is spent.
• Detect N+1 queries: Check the call count on edges between a service and a database to spot repeated queries.
• Drill into a service: Click an erroring or anomalous node to view its entry spans and errors, then jump to those spans in the waterfall.

How services are displayed:

If a service is called by more than one parent service, it appears as a separate node under each caller. This keeps the map free of cycles and shows exactly which path each call took.

Use custom Datasets in the Trace Explorer

Observe by Snowflake's trace data features work best with the Datasets from the Tracing/ package, which are available when you configure data ingestion using the Add Data portal.

If you want to use a custom span Dataset, here’s what you’ll need to provide in the Dataset definition (Observe by Snowflake automatically builds traces from spans, so you don’t need to provide a trace Dataset):

• Interval dataset (validFrom / validTo) with start_time
• end_time
• trace_id
• span_id
• span_name
• response_status

The custom span Dataset requires a primary key on trace_id and span_id . To do this, add the following to your Dataset definition:

set_primary_key trace_id, span_id

Don't add llm_span_type, that routes the Dataset to LLM Explorer. See LLM observability.