Skip to content

Warehouses

A warehouse is where Aunsight Data Services stores structured data — tables of rows and columns that can be queried with SQL. Warehouses are the backbone of the structured data pipeline: acquired data lands in a warehouse, is mapped and processed through further warehouses, and is ultimately delivered from one.

Warehouse types

ADS works with warehouses in a few forms, depending on where the data lives and what it's used for.

Local Warehouse

A Local Warehouse is a warehouse managed inside ADS, within a compute environment. This is the primary structured store for a project — it's where acquired source data lands and where the models built on top of it live. A single project can contain one or more local warehouses.

The models in a solution project are all local warehouses at different stages of the pipeline:

  • the Local Source Model — acquired source data, as it first lands in ADS
  • the Mapped, Processed, and Deliverable models — successive transformations of that data toward the solution's output

See Local Source Model for how acquired data appears and is explored in a project, and Local Warehouse for the full warehouse and table interface (settings, schema, exploring data, and ingest).

Remote Warehouse

A Remote Warehouse is a warehouse in an external source system — a client's database, for example — that ADS reads from during acquisition. It sits outside the platform; data is pulled from it into a local warehouse through a connection and an ingestion task. The remote warehouse (or bucket) as it exists at the source is sometimes referred to as the remote source model.

User Warehouse

A User Warehouse is a warehouse scoped to an individual user or to a pipeline module's working space, rather than to the shared project pipeline. It's useful for staging, intermediate results, or user-specific outputs that don't belong in the main project warehouses. Data can be transferred between user warehouses and local warehouses as part of a pipeline.

Accelerated Warehouse

An Accelerated Warehouse (AWHS) is a high-performance warehouse engine used for fast queries and for AI-oriented features such as vector (semantic similarity) search. Because it's primarily part of how data is consumed — including through the Consumption API and agent platforms — it's covered on the consumption side rather than here. See Data delivery & consumption for details.

How warehouses fit together

In a typical solution project, structured data flows through a series of warehouses:

  1. Data is acquired from a remote warehouse (the source) through a connection and ingestion task.
  2. It lands in a local warehouse as the Local Source Model.
  3. It's mapped and processed through further local warehouses (mapped, processed, deliverable models).
  4. It's delivered or consumed — optionally via an accelerated warehouse for high-performance or AI-driven access.

User warehouses sit alongside this flow as scratch or output space for individual users and modules.