Changelog

New

• Added an interactive month carousel to the SLA and SLO performance over time charts, allowing users to focus on the most recent months

Updates

• Added autocomplete suggestions to multi-metric SELECT queries in the SLI definition flow, displaying available metrics as you type to make metric selection faster and more accurate

  

Fixes

• Fixed how we display SLO breaches in the calendar view to better reflect their actual status

  • Previously, when an SLO went into breach (e.g. on Sept 5th), subsequent days would incorrectly show as orange/warning even though the SLO was still in breach and performing below target

    

  • Now, all days where a breached SLO performs below its daily target will correctly show as red/critical (ex Sept 27th), providing a more accurate visualization of the breach's duration

    

New

• Freeform SELECT clause in SLI Builder: Users can now create SLIs using free-form SELECT clauses with enhanced SQL query handling. This includes support for arithmetic relationships between multiple metrics in the SELECT clause.
  Example: SELECT AVG(internal_timestamp - external_timestamp) FROM key1 WHERE status = "COMPLETED".
  

  Note: Aggregation functions are required in the SELECT clause for complex relational queries.

Updates

• Breach start time for interval based SLOs: We’ve introduced the breach_starts_at field for interval-based SLOs to enhance breach tracking. This addition provides greater context on the SLO’s state over its compliance period. You can view the breach start time by clicking the </> inspect icon on the SLA details view.

Fixes

• SLA calculation for mid-month starts: We resolved an issue causing compute errors when SLAs began mid-month.

• Improved data handling for interval exclusions: Charts now correctly show "no data" for interval-based SLOs where missing intervals are set to "exclude."

Updates

• Clickable SLA rows: You can now click anywhere on an SLA row or card to access detailed information. No more hunting for that tiny arrow icon – just click and go!

  

• SLO counters: We've added a counter showing the number of SLOs for each SLA in the Organization page to help you quickly assess the complexity and scope of your SLAs without needing to drill down into each one

  

• Better charting for in-progress days: Computations for the current day are now represented by a dashed line in charts to tell you that data for the day is still being collected and may change. We hope it helps prevent misinterpretation of potentially incomplete information. At the end of day, the chart line will turn solid.

• Performance boost with batch processing: We’ve implemented batch processing for OpenSearch writes and event generation.This behind-the-scenes change should result in noticeably faster data processing and reduced system load. You'll see quicker updates and smoother performance, especially when dealing with large volumes of data

• Improved table sorting: We’ve updated default sorting for several key views:

  • SLOs are now sorted alphabetically by name. Alphabetical SLO sorting addresses the scenario with many objectives where you want to easily compare the configuration of Objectives with different targets (example P90 Latency and P99 Latency).

  • SLAs are now ordered by end date, putting your most pressing agreements front and center.

• Proactive data freshness: We’ve implemented automatic recomputation for SLAs not processed in the last 24 hours to ensure your SLA data is always up-to-date, even for less frequently accessed agreements.

• Static SLA portal links: The SLA portal links are now updated to remain static upon deactivation and reactivation thereafter so that your active integrations don’t need to be updated.

Fixes

• Error budget for in-progress day: Error budgets for the current day were incorrectly showing as zero when no data was available. We've corrected the client-side logic to properly display the error budget, even when today's data hasn't arrived yet to prevent false alarms and confusion.

• Improved timezone handling: Some timestamps were being processed inconsistently across the system depending on the end user’s timezone. We've standardized timestamp handling, including making our event generator timezone-aware (defaulting to UTC). This fix ensures your data is correctly time-aligned and your SLAs are evaluated appropriately.

• Correct handling of computation based on SLA policy period: Our SLA computation logic was not correctly aligning with the official start date of SLA policy periods. Instead, it was basing calculations on the first ingested data point for each SLO. We've updated our computation logic to properly account for the SLA's official start time, ensuring all relevant events/intervals within the policy period are included in SLA status calculations.