Data Connections

Data Connections

What is a Connection?

A Connection is a secure link between Spartera and your data source.
You can set up private (user only) or shared (team) connections using
secure authentication with service accounts, API keys, and bearer tokens.

How does Spartera connect to my data?

Spartera supports multiple backends including:

• Cloud Data Warehouses: Google BigQuery, AWS Redshift, Snowflake,
  Teradata Vantage, Databricks SQL, Supabase Analytics.

• Databases: Cloud SQL (MySQL/PostgreSQL), RDS MySQL, Aurora
  (Serverless & Provisioned), SQL Server, Supabase PostgreSQL.

• Model Serving & External APIs: Databricks Model Serving, and any
  HTTPS API via an External / Data API connection — your own ML model
  endpoint, or a third-party data API you package and resell.

Each platform requires specific authentication methods (see the table
in the UI for details), such as JSON key files, API keys, bearer tokens,
or database credentials.

What can I do with an External / Data API connection?

External API connections serve two purposes. You choose which when you
create the connection, using the API Type field:

• Managed Calculation — Spartera calls your API to compute a single
  value (for example, an ML model prediction) and sells it as a
  Calculation asset. Your model, training data, and logic stay on
  your own infrastructure.

• Data Feed Passthrough — Spartera calls a data API, repackages the
  response into the rows and fields you configure, and resells it as a
  Data endpoint. Ideal for value-added reselling of third-party
  data APIs.

Your API Type determines what the connection can back: Managed
Calculation connections power assets; Data Feed Passthrough connections
power endpoints.

How secure are our credentials?

Connection credentials are stored in Google Cloud Secret Manager, which
encrypts them at rest. They are never returned in API responses, and you
can rotate or remove them at any time.

Do you store or move our data?

Spartera never stores your raw datasets, tables, or files, and keeps no
copy of them. Only what you choose to publish leaves your systems. What
happens depends on the product:

• Assets (calculations & models) — Spartera runs your analytic logic
  against your source, or calls your API, and returns only the computed
  result. Your raw data never leaves your infrastructure.

• Endpoints (data feeds) — When you publish an endpoint, you are
  deliberately choosing to deliver data to buyers. You stay in full
  control of exactly what is shared: which fields are exposed, the format
  they are returned in, and the permissions and rate limits that apply.
  Data is served live at request time. Spartera keeps only a sample
  preview (three records by default, generated when you preview your own
  listing) and the snapshot a buyer purchases, which is held for that
  buyer's account only. The full list of what Spartera retains is in
  Section 9 of the Marketplace Terms.

In every case, you — the data producer — decide what is shared and how.
Nothing is moved, delivered, or exposed without your configuration.