Query Analysis¶
Top queries¶
PerfScope's primary analysis view surfaces the Top N worst-performing queries for the selected time range.
Queries can be ranked by metrics such as:
| Metric | What it helps identify |
|---|---|
| Execution Time | Long-running queries |
| Credits / compute cost | Expensive workloads |
| Bytes Scanned | Queries processing unusually large amounts of data |
| Spillage | Queries exceeding available memory and spilling to storage |
| Queuing Time | Workloads waiting for warehouse resources |
Select a ranking metric to reorder the results.
Drill-down analysis¶
Select a query to inspect available diagnostic context, including:
- Warehouse
- User
- Role
- Query tag
- Start and execution timing
- Scan characteristics
- Spillage indicators
- Query-performance metadata
The exact fields displayed can vary based on the metadata available from Snowflake.
Trend analysis¶
Use trend views to understand how workload behavior changes over time.
Trend analysis can help you:
- Detect performance regressions
- Compare different time periods
- Identify recurring workload patterns
- Spot increases in query cost
- Identify warehouses experiencing growing resource pressure
Filters¶
Use filters to narrow the analysis.
| Filter | Description |
|---|---|
| Warehouse | Focus on a specific warehouse |
| User | View queries executed by a specific user |
| Role | Filter by execution role |
| Query Tag | Filter by application, pipeline, or workload tags |
| Time Range | Adjust the analysis window |
| Query Type | Filter SELECT, INSERT, MERGE, COPY, and other operations |
Recommended workflow¶
A practical PerfScope workflow is:
- Select an analysis time range.
- Rank queries by execution time or compute cost.
- Inspect the top offenders.
- Filter by warehouse, user, role, or query tag.
- Check spillage and queuing indicators.
- Compare with a previous time period.
- Use the findings to prioritize query or warehouse optimization.