Workspace configuration

On the workspace settings page you can set cluster defaults, create labels, add service accounts and change other workspace related settings.

Workspace configuration

Configure workspace identity through the name and description fields. Use descriptive naming conventions that align with your project structure or data domains for easier workspace discovery and management.

Default Compute Configuration

Set default computational resources to standardize job execution without requiring explicit compute context configuration for each operation. Two configuration modes are available:

Resource-based configuration allows precise specification of vCPUs, RAM, storage, and cluster size. Use this approach when you require control over resource allocation or have specific performance/cost optimization requirements. When the specific configuration is not available on AWS, the cheapest instance with at least the specifications is selected and used.

Instance-based configuration leverages cloud provider instance types. This approach simplifies configuration by using pre-optimized instance configurations and is typically more cost-effective for standard workloads.

Default compute configurations eliminate the need to explicitly define a compute context. More information on configuration can be found in the section on setting the compute context.

Query and compute labels

Labels help organize and categorize queries and compute within your workspace.

Labels on a query are created the first time you use them, so you can pass any name:

lf.remote(ctx).labels(["docs", "user-guide"]).sink_parquet("s3://bucket/output.parquet")

A label name must be 1-32 bytes long, without control characters or surrounding whitespace, and a query can carry at most 64 labels.

Labels on compute must already exist in the workspace, and are created and edited through the dashboard interface:

ctx= plc.ComputeContext(
    workspace="your-workspace",
    labels=["docs", "user-guide"],
)

Service Account Management

Service accounts allow scripts to authenticate without human intervention. Each service account generates an authentication token with workspace-scoped permissions.

Use clear and descriptive names that match your project structure or data domain to improve clarity. Refer to the Use service accounts section for implementation patterns and security best practices.

Disable workspace

Workspace admins can disable workspaces removing all data, terminating active processes, and revoking all access tokens. This action cannot be reversed.

Ensure data export and backup completion before disabling workspaces containing critical datasets or analysis artifacts.