Ask or search…
⌃K
💡

Release Intelligence

Quickly analyze your releases to understand changes in monthly cloud spend, latency and errors.
Release Intelligence empowers your team to build and iterate faster by analyzing release performance on your behalf so that you can quickly review results and intervene as needed. Sedai summarizes a release's performance so you can validate a release's improvement or troubleshoot unexpected behavior.
Sedai automatically detects new releases for your compute resources and analyzes latency and error metrics under varying traffic levels. Each release is graded based on its performance quality relative to the previous release.
Scores reflect how much a release deviates from its prior release, and can indicate improvements or degradation in monthly cloud spend, latency and/or errors. If a release's score is negative, we recommend reviewing your release's performance closley to avoid service disruptions.
Release Intelligence is automatically enabled for all integrated cloud accounts. If you don't want to receive scorecards for specific resources, you can disable the feature for an account, group, or per resource from the Topology Settings page.

About Scores

Sedai analyzes a resource's performance metrics pre-release and creates a model that predicts behavior post-release. Optimal metrics are automatically selected and analyzed. These metrics are used to calculated individual scores for latency and errors, and saturation metrics are used to estimate a release's monthly cloud spend. All three scores are aggregated into a summary score to indicate positive or negative deviation compared to the previous release.
Analysis is based on datasets pulled from different time windows and varying levels of traffic (in case of insufficient traffic post-release, Sedai may be unable to generate a score).
The summary scores is based on a scale of -10 to +10. Low scores indicate a negative impact, while positive scores reflect improvement. If a score is close to zero, it merely indicates the new release has not changed much from its previous release.
Keep in mind that Sedai grades each resource individually based on its respective history. A score could indicate a significant increase in monthly cost, errors or latency, but the results may be considered reasonable. We recommend reviewing details in a scorecard's side drawer to better understand the analysis and determine whether the release is behaving as expected.
A preliminary score is generated shortly after a release is detected, but a score may be updated during the ongoing analysis up to 24 hours post-release.
If there is insufficient traffic for analysis, the score will be marked as unavailable (N/A).
Only the latest release will be analyzed. If a new release is detected while an analysis is already underway, the existing analysis will be completed as-is and a new analysis will automatically be triggered.

View Scorecards

Navigate to Release Intelligence from the side navigation. Here you will find a list of all connected cloud accounts or Kubernetes clusters and an overview of estimated monthly cost, average latency, and errors for all resources within it, as well as a summary of total releases and their average score for the selected time period.
Within an account or cluster you can view a breakdown of scorecards by their release date to quickly understand your overall release trends. You can filter by the summary score or by the individual scores for cost, latency or errors for a more granular view. You can further refine this view by selecting a different time period or overlaying filters for groups or specific resource types.
Select a day to view a list of scorecards. You can filter the list by the summary grade as well as by release type (if it included code and/or configuration changes).
If the selected day included releases with autoscaling events, you can optionally reveal scorecards for them. By default, this option is deselected.
Select a scorecard to view more details about the score itself as well as see configuration details and a history of previous releases for the resource. You can also see an overview of release trends to better understand how a resource's cost, traffic, latency and errors have changed over time.
​
Release Intelligence is supported for the following resource types:
  • AWS Lambda
  • AWS ECS
  • Kubernetes (self-hosted or managed)
Last modified 20d ago