πŸ”‘
Cloud Access
Learn about Sedai's access permissions and how they are used to autonomously manage your cloud resources.
At a basic level, Sedai requires cloud access to discover your topology. Additionally, Sedai needs read-write access in order to act on your behalf if it detects opportunities to improve availability, performance, and/or cost. This also enables Sedai to learn from its actions and continuously improve its ML models so that it can intelligently adapt to your cloud resources.
While the platform needs read-write access in order to train its algorithms, you can configure how Sedai manages your resources from the platform's Settings page. This allows you to control:
  • What Sedai can act upon: Permissions for an account or group (including defaults as well as overrides at the resource level)
  • Why Sedai acts: Feature enablement, such as whether Sedai looks for opportunities to improve availability or optimize for cost/performance
  • How Sedai acts: The types of actions you want Sedai to autonomously execute
  • When Sedai acts: Permission to act independently as it identifies opportunities, or alert you and ask for permission to act
With these settings you can optionally receive recommendations as you get started with Sedai, and then seamlessly transition to autonomous mode when you're ready.
Sedai will only act autonomously once learning is complete and the platform is confident it can act safely without harming resource behavior. You can also manage which resources Sedai autonomously manages from the Settings page.
Even if you do not intend to permit Sedai to autonomously act on your behalf upfront, we recommend configuring connected clouds with Sedai's autonomous permissions so that you retain the option to enable the platform's learning loop when you're ready.
Alternatively, you can attach the use Sedai's read-only permissions, which grants Sedai limited access to observe your cloud. Sedai will only generate low-confidence recommendations such as irregular resource behavior as well as potential opportunities for performance improvements. However, since Sedai cannot act, it is unable to learn from its observations to improve future recommendations.
View what's included in Sedai's policies:

The following cloud providers leverage either Sedai's autonomous or read-only policies with IAM authentication. Select a cloud to view what is included in its respective policies:
For Kubernetes-based resources, Sedai can also connect to clusters via its Smart Agent.
In addition to cloud access, Sedai also requires access to monitoring data:
Copy link