Analyzing Query Success, Failures, and Performance Using the FinishQuery Table

Summary: mcmhu6 provides an overview of Sisense's out-of-the-box 'Usage – Query' dashboards designed to analyze query performance. The article introduces two dashboards, one for successful queries and another for failed queries, aimed at helping admins optimize performance by identifying bottlenecks and troubleshooting failures. Implementation steps include enabling usage analytics, creating custom tables, and importing the dashboards. m58fqu asks for the referenced files, and mcmhu6 confirms they have been uploaded, inviting feedback on any issues during implementation. The solution is custom and may require adaptations for different environments.

Background

Sisense provides an out of the box Usage – Query dashboard when the Usage Analytics model is enabled. This dashboard offers insights into queries created and executed by users.

The goal of this article is to introduce a structured flow for Admins to review overall query performance and drill down into successful and failed queries. To support this analysis, the solution is divided into two dashboards, one focused on successful queries and one on failed queries, enabling efficient troubleshooting and performance optimization.

The Dashboards

Successful Query Performance Dashboard
This dashboard analyzes successful queries only and helps Admins evaluate overall query performance by:

Identifying performance bottlenecks and understanding when they occur
Analyzing which execution phase (translation, data source execution, throttling) contributes most to total query duration
Highlighting whether the source of performance impact is clearly identified or requires further investigation
Reviewing query performance at the dashboard level, widget level, and individual query level
Tracking query behavior and performance over time, including date and time-based patterns to identify peak hours, recurring slowdowns, or workload-related trends

For best results, it is recommended to:

Review the ReadMe widget included in the dashboard before getting started, to gain a better understanding of the data, definitions, and assumptions used in the analysis
Analyze the dashboard by asset query source, using single-select filtering for each option to ensure focused and accurate analysis:
- Dashboard queries
- Widget queries
- Compose SDK queries
- API / JAQL queries
Filter by duration to review longer-running queries and understand the factors contributing to their performance.
Review performance by data model source, also using single-select filtering to isolate behavior and performance patterns:
- Live models
- Cube-based models

Failed Queries Performance Dashboard

This dashboard analyzes failed queries only and helps Admins investigate and understand query failures by:

Identifying the root causes of failed queries using exception messages and categorized error types
Pinpointing when failures occur, including exact date and time, to detect recurring patterns or peak failure windows
Analyzing failures by data source and data model, helping determine whether issues are related to Live connections, Cubes, or specific providers
Reviewing failed queries at the dashboard level, widget level, and individual query (Query GUID) level, enabling precise troubleshooting for specific widgets or assets
Filter Out Canceled queries

Implementation Steps

Unzip the provided file locally.
The ZIP file contains:
- A TXT file with the SQL used to create the Fact_finish_query table
- The Successful Query Performance dashboard file
- The Failed Queries Overview dashboard file
Enable Usage Analytics (if not already enabled):
- Navigate to Admin → Monitoring & Support → Usage Analytics
- Enable Collect Usage Analytics
Open the Usage Analytics model:
- Navigate to the Data tab
- Open the Usage Analytics data model
Create a custom table:
- Add a new custom table named Fact_finish_query
- Paste the SQL from the provided TXT file into the table definition
- Parse the SQL Statement by Preview the table
- If Parse Successfully --> Click Save
- Build the data model:
  - Run Build (Full or Changes Only)
- Import the dashboards:
  - Navigate to the Analytics tab
  - Import both dashboards from the unzipped files

Implementation Troubleshooting
If the Custom Table import does not parse successfully, the issue is most likely related to columns that are in a dropped state.

Below is an example error.
The import fails because the Custom Table references a column that is not declared in the FinishQuery table.

Steps to review

In the Table Columns left panel, search for the FinishQuery table.

Scroll down to the dropped columns at the bottom of the column list, which are highlighted inline.

Open the column menu and click Restore.

Add the Custom Table and verify that it was parsed successfully.

Build the data model:
- Run Build (Full or Changes Only)
Import the dashboards:
- Navigate to the Analytics tab
- Import both dashboards from the unzipped files

Disclaimer

This is a custom solution provided for reference only. It may require modifications to work in different environments and could be affected by future product updates. Customers are responsible for testing and validating the solution before using it in production.