Debug Web UI for Query Timeline

Analyzing query performance by using the timeline

For a detailed report on how a query was run and to understand its performance characteristics, you can use the Gantt chart in the built-in WebUI. This chart is an alternative to the profile command and renders timing information and dependencies graphically.

The Gantt chart supports the following analysis capabilities:

Determining time consumption — Use the chart to identify where queries are spending time instead of searching through text-based profiles.
Drilling down: After you identify which fragments or operators consume the most time, you might need to check the summary or full profile for more detail.

The following image shows the Gantt chart for the query timeline of a TPC-DS query:

Figure 1. Query timeline in Impala WebUI

Visualizing fragment and node execution

The query timeline provides a visual representation of query fragments and plan node timings in a compact format.

- Tree visualization: The chart displays the plan and fragment tree on the left with corresponding timing bars on the right.
- Color coding — Plan node labels are color-coded by fragment for easy identification.
- Phase tracking — Each colored block represents the longest time spent on a task. Tasks that complete earlier are displayed as grey lines.
  Figure 2. Visual representation of query fragments
- Join and exchange indicators — Dotted boxes for exchange nodes display when data rows are sent, while specific symbols indicate when joined data arrives.
- Flexible ordering: You can use the checkbox to change the node order to match the plan order used in the Summary tab.
- Automatic updates: If a query is still running, the timeline reloads every second to provide the latest progress updates.

Correlating resource and utilization metrics

The WebUI provides detailed resource metrics to help you identify hardware bottlenecks directly from the timeline.

CPU utilization tracking

You can view a stacked area chart that displays aggregate CPU performance across different nodes throughout a query.

Metric types: The chart tracks host CPU utilization for user, system, and iowait percentages.
Data alignment: Utilization samples align with the fragment timing diagram to help you correlate resource usage with specific query parts.
Resource tracing: The RESOURCE_TRACE_RATIO query option determines the utilization values tracked within the runtime profile.

Memory, disk, and network throughput

Fragment-level usage: Click any plan node to view the total memory and thread usage for that specific fragment.
Host throughput: When resource tracing is enabled (using query option RESOURCE_TRACE_RATIO=1), the timeline displays real-time disk read and write speeds and network transmission rates.
Synchronized inspection: Hovering over the diagram displays a time-synced line across all charts, with a tooltip showing exact timestamps and metric values.
Readable units: Memory values automatically scale to clear units such as megabytes or gigabytes for better readability.

The following image shows the CPU utilization and resource metrics chart in the Impala WebUI:

Identifying instance-level skew

The WebUI helps you identify "skew," which occurs when specific tasks take significantly longer than others, by using a histogram display.

Distribution histograms: For phases with more than five instances, the timeline groups timestamps into divisions.
Visual density: The height of each rectangle shows how many instances fall into that time range, while their width shows the maximum timestamp in each division making it easy to spot outliers.
Skew tooltips: Hovering over a phase rectangle displays the number of tasks in that bucket along with the minimum, maximum, and average time taken.

Example:

When a phase contains more than five instances, the timestamps are bucketed into five divisions. Each division spans 20% of the difference between the maximum and minimum timestamp values. The height of each rectangle is proportional to the number of instances in that division. Hovering over a bucketed phase rectangle displays a tooltip that contains the following details:

Number of instances
Maximum timestamp of the event in seconds
Minimum timestamp of the event in seconds
Average timestamp of the event in seconds

Additionally, this update minimizes the generation of SVG elements, which significantly decreases browser rendering times for large query profiles.

You can use the following shortcuts while hovering over the timestamps list:

To increase the number of timestamps: Hold Shift + Mouse Wheel while hovering on the timestamps list.
To increase precision of all numbers in query timeline: Hold Alt + Shift + Mouse Wheel while hovering on the timestamps list.

The following histogram chart shows low instance-level skew across divisions:

The following histogram chart shows moderate variance and outlier distribution in query execution:

The following histogram chart shows high instance-level skew with severe processing outliers:

Managing and exporting query data

You can export query data for offline analysis or import historical profiles to troubleshoot past issues.

Visual exports: You can export query plans and timelines as vectorized image files(SVGs embedded within the HTML). These files include the unique query ID and allow you to scale the diagram without losing quality.
Data imports: You can select and upload multiple profile files from your local computer at once.
Local storage: Imported profiles are stored in your browser, allowing you to review a large number of historical queries without re-uploading them.
Structured profile view: The query profile tab converts imported data into an organized text format for better readability.

The following image shows the initial profile selection menu for data imports:

The following image shows the multi-file import progress tracking status:

The following image shows the visualization of an imported historical query profile:

Performance and interface maintenance

The WebUI has been updated to use modern web standards to improve performance and reliability.

Memory efficiency: Updated script standards reduce browser memory usage and prevent technical conflicts within the interface.
Standardized naming: All internal labels and functions follow a consistent naming pattern to make the interface easier to maintain and update.
Optimized rendering: The system groups multiple tasks together to significantly speed up loading and rendering times, even for very large and complex queries.