Get started
The shortest path from a new workspace to a query, a data sync and traceable lineage.
There is no useful “hello world” for a cloud data and AI platform if it avoids your real system. A better first pass is to connect one source, move one table, run a known query and attach the business context that makes the result useful.
Before you begin
You need a Vegalake organization, a workspace and permission to create resources in that workspace. Administrators can keep production and development in separate workspaces, or use VegaGraph contexts when the same logical estate needs environment-specific facts.
Use organizations and workspaces to prepare your scope and access management to assign permissions. Set up a Secret Vault before creating a connection that needs credentials.
Create compute
Create a small VegaFlow cluster for ingestion and a VegaDB warehouse for SQL. Compute is isolated by workload, so a large backfill does not need to compete with an analyst's query.
Add a connection
Add a supported source in VegaFlow and validate its configuration. Secrets are stored as managed secret references; they are not returned in read responses.
Run a QuickFlow
Discover the source catalog, select a stream, choose the destination behavior, and run it once before adding a schedule. The run page exposes stream attempts rather than hiding partial failure behind one status.
Query the result
Connect to the VegaDB endpoint using psql, pgJDBC, psycopg or node-postgres and run the same SQL you intend to put in production.
Attach context
Find the table in VegaGraph, add ownership and business definition aspects, then inspect its upstream and downstream paths. Follow the dashboard walkthrough to explore an existing resource.
What to do next
- Read VegaDB clusters before sizing production workloads.
- Read QuickFlow run behavior before setting a frequent schedule.
- Read contexts and lineage before representing development, staging and production.