Workgroups & Workspaces

In TigerGraph Savanna, resources are organized into a clear hierarchy to manage access and cost efficiently.

1. The Resource Hierarchy

Understanding the relationship between these three entities is critical:

Level	Name	Purpose
L1	Workgroup	Defines the "Home" for your projects (Cloud Provider + Region).
L2	Workspace	The compute cluster. Defines the CPU/RAM and runtime version.
L3	Database	The persistent graph data (Schema + Vertices + Edges).

2. Workspaces: RW vs. RO

Because compute is separate from storage, Savanna introduces two types of workspaces:

• Read-Write (RW): The primary workspace used for schema changes and data loading. Only one RW workspace can exist per database.
• Read-Only (RO): Secondary workspaces used for querying. You can have multiple RO workspaces to handle high-concurrency read traffic without impacting the RW instance.

3. Dynamic Scaling

• Expansion & Shrink: You can adjust the size of a workspace (e.g., from Small to X-Large) as your computational needs change.
• Scheduling: Workspaces can be scheduled to automatically start and stop, ensuring you don't pay for idle compute time.

4. Workgroup Management

A Workgroup acts as a security boundary. Within a workgroup, you can:

• Invite team members.
• Manage IAM policies.
• Monitor total resource usage across all child workspaces.

[!IMPORTANT] If you terminate a Read-Write workspace, you cannot re-create it for that specific database. Ensure you have backups or a clear migration path.