Core Concepts
Organization and identity
An organization is your company's or team's space in Gnok. Its users, data, roles, and settings are kept separate from every other organization. SQL system views and APIs call an organization a tenant (for example, a tenant_id column).
You sign in with your organization's sign-in name (such as acme_corp), your email, and your password, plus a second factor when your account has one. See account access. Your username is the name Gnok uses for you inside the organization, for example in grants and query history.
Roles and grants determine which data and administrative operations you can access. A Studio administrator's navigation options do not replace the engine's object-level permission checks.
Studio connection
A connection tells Studio which Gnok endpoint to use and holds defaults such as catalog, schema, and warehouse. The provisioned Gnok connection uses your signed-in identity. A shared connection gives users a common endpoint; it does not give them another user's privileges.
Catalog, schema, and table
Objects are addressed as catalog.schema.object, for example tutorial.orders_demo.orders. A catalog groups namespaces and table metadata; a schema groups related tables and other objects. Session defaults let you shorten names, but qualified names make reusable SQL easier to understand.
Gnok uses Apache Iceberg for table metadata and snapshots. A snapshot identifies a committed table state. Time travel reads an earlier state when the relevant snapshots and files are retained.
Compute warehouse
A compute warehouse is the execution resource associated with a workload. Use the warehouse made available to your account when one is required. Its size, state, and usage affect execution and consumption. Warehouse administration requires permission.
The word warehouse also appears in Iceberg catalog client configuration, where it can identify a catalog or storage location. Use the meaning documented by that interface; it is not always a compute warehouse name.
Worksheet and saved query
An editor tab is a working draft. A saved query gives SQL a name so you can reopen or share it. Shared tutorials are saved SQL templates. They execute through your selected connection and require your own data privileges.
Models, indexes, and procedures
A model defines the input and output contract for training or inference. A vector index supports similarity search over a vector column. A procedure packages SQL statements and control flow for invocation with CALL. These are separate objects with their own lifecycle; saving a worksheet does not create them until its SQL executes.
Interactive and background work
A query runs with the submitting user's identity. Scheduled or background features may use a separate service identity scoped to your organization. Such an identity needs explicit grants for its source and output objects. See AI and ML prerequisites.