You know the drill. Someone on the data team needs quick read access to CosmosDB to validate production metrics. Another engineer wants to run a reproducible experiment in Domino Data Lab using that same dataset. Nobody wants to juggle secrets or permissions at 2 a.m. Getting these systems to trust each other without leaking credentials is the real puzzle.
CosmosDB brings globally distributed storage with instant scalability and multi-region replication that keeps latency low. Domino Data Lab delivers secure, centralized workspaces for data science teams to build, train, and deploy models repeatably. When you connect them right, the result is faster iteration with guardrails in place. The key is identity. You need policies that travel between systems instead of static blobs in environment variables.
The CosmosDB Domino Data Lab integration workflow aligns identity with data access. Domino authenticates users through an identity provider like Okta or Azure AD, then requests temporary credentials for CosmosDB through managed service principals. That exchange combines OIDC tokens for the user’s session with RBAC controls already defined in Azure. No shared credentials, no long-lived keys hiding in notebooks. Everything traces back to a real person and a real policy.
To make this repeatable, treat permissions as infrastructure. Use Terraform or your preferred IaC tool to define which Domino projects can touch which CosmosDB containers. Rotate keys automatically through Key Vault and map each role explicitly. If an access check fails, Domino handles the error gracefully—no silent retries that hide the mismatch. Audit logs then tie every data read back to the source identity. It’s clean, measurable, and meets SOC 2 expectations without drama.
Benefits of integrating CosmosDB with Domino Data Lab: