In certain cases, like extremely high volumes or if a team generally only queries a subset of the data without requiring wider correlation, running queries with filters on the entire index can be slow. These scenarios benefit from making the initial query space smaller itself.
With Physical Indexes, you can do exactly that. Define the filters, give the new index a helpful name, and choose whether to keep the same data in the main default index. For some of our largest customers, queries over long durations have dropped to sub-second performance.
Read our docs for more details. Physical Indexes are currently being enabled on request depending on requirements. If you think it might be helpful for you, please reach out to support.
Improvements
- Logs
- Add selected body text or an attribute value as a negative filter to the current query (does not contain, does not match, is not empty)
- Link to Trace Details when trace_id is available in attributes or JSON body of a log line, improving debugging workflows
- Improved handling for duplicate filters and filter values
- Improved support for prettifying JSON in Tail mode
- Improved support for OpenTelemetry semantic conventions for displaying, searching, and filtering attributes
- Dashboards
- View deployment events, feature flags, and other change events in line charts
- Alerting
- Improved handling of repeated variables in alert annotation templates
Fixes
- Fixed issue where currently selected log line was not shown in related logs while viewing log details in some cases