In the recent releases, we have built the backend framework and APIs for query insights, including the capability to identify the top N queries based on various metrics. However, users currently lack a dashboard tool to visualize and analyze this data. They can only query the top N queries via an API and must manually aggregate and interpret historical data on their own for visualization purposes. This process is cumbersome and limits the ability to quickly diagnose and resolve performance issues.
What solution would you like?
Develop the MVP verison of the query-insights-dashboards plugin, with the best practices mentioned in the dashboard plugin development guide.
Integrate the dashboard with the top n queries API, to develop the overview page on point-in-time top n queries in the current cluster. Fill any gaps in the API (if there’s any).
Integrate with the top n queries exporter, to display historical top n queries on the overview page. Fill any gaps in the exporter (if there’s any).
Develop the drill-down view for the top queries, for detailed query analysis.
Develop the configuration view to more easily configure the query insights plugin.
Mock screens
Screen 1: Top N queries overview page. Users will be able to see the top queries
Screen 2: Query details screen. Users can click on the time stamp to drill down and view the details of the query.
Screen 3: Configuration tab displays the default configuration settings. Users with appropriate edit access can modify the configuration settings.
Screen 3a. While an edit is being made, the bottom bar is displayed with actions to save/cancel, along with the status of number of unsaved changes.
Screen 3b. Updated configuration is saved once user clicks “Save“.
Is your feature request related to a problem?
In the recent releases, we have built the backend framework and APIs for query insights, including the capability to identify the top N queries based on various metrics. However, users currently lack a dashboard tool to visualize and analyze this data. They can only query the top N queries via an API and must manually aggregate and interpret historical data on their own for visualization purposes. This process is cumbersome and limits the ability to quickly diagnose and resolve performance issues.
What solution would you like?
Mock screens
Screen 1: Top N queries overview page. Users will be able to see the top queries Screen 2: Query details screen. Users can click on the time stamp to drill down and view the details of the query. Screen 3: Configuration tab displays the default configuration settings. Users with appropriate edit access can modify the configuration settings. Screen 3a. While an edit is being made, the bottom bar is displayed with actions to save/cancel, along with the status of number of unsaved changes. Screen 3b. Updated configuration is saved once user clicks “Save“.
What alternatives have you considered?
N/A
Do you have any additional context?
https://github.com/opensearch-project/OpenSearch/issues/11522