Skip to main content

SaaS Deployment

Overview​

SaaS is the default OpsWorker deployment model. OpsWorker runs as a fully managed, multi-tenant platform on AWS. You install only the read-only Kubernetes Agent in your clusters; everything else is run and maintained by OpsWorker.

How It Works​

ComponentManaged By
Investigation engineOpsWorker
Data storageOpsWorker
PortalOpsWorker
AI models (AWS Bedrock)OpsWorker
Updates and maintenanceOpsWorker
Kubernetes AgentYou (Helm install)

The agent is read-only (it uses the built-in Kubernetes view role), outbound-only, and communicates with OpsWorker over SQS using an assume-role credential. A default install runs three pods: the agent, the kubernetes-mcp-server, and the grafana-mcp server. See the agent documentation for details.

What You Get​

  • Low operational overhead. No OpsWorker infrastructure to provision or maintain.
  • Continuous updates. New features and fixes are deployed by OpsWorker without action on your side.
  • Scales with your workload. The serverless backend scales automatically.
  • Quick setup. Sign up and connect your first cluster in roughly 10 minutes.

Multi-Tenancy​

SaaS is multi-tenant with organization-level data isolation. Your data is partitioned and encrypted, with no cross-organization access. See Data Isolation.

AI Models​

Investigations run on AWS Bedrock (Amazon Nova for fast extraction and routing, Claude for reasoning), with the direct Anthropic API as an alternative reasoning provider. OpsWorker manages all model access; there is no configuration required on your side.

Best For​

  • Teams who want the fastest setup with no infrastructure to manage
  • Organizations comfortable with managed SaaS data handling
  • Most OpsWorker customers

Next Steps​