Data warehouse

Data warehouse lives in the Settings area of the dashboard, under Platform.

At a glance#

What it does#

Warehouse export is a continuous pipe from ScaleBun into your own data warehouse. You configure a destination, see its status, run it on demand, and remove it when it is no longer needed.

Unlike Data export, which produces a file once, this keeps delivering.

When to use it#

When ScaleBun's data needs to sit alongside data it does not have. Product analytics in isolation cannot answer questions that require your billing system, your support tickets, or your inventory — and no amount of dashboard features will change that. Joining in your own warehouse can.

Also when you need a query language. The dashboard's analysis is deliberately bounded; SQL is not.

Workflow#

1. 1

   Model on a small range before syncing everything

   Confirm the schema is what you expect and that your transformations work. Backfilling a mistake is expensive.

2. 2

   Run on demand first

   Prove the connection and the credentials before relying on the schedule.

3. 3

   Watch the status after the first few runs

   A pipeline that silently stopped is worse than no pipeline, because everything downstream keeps returning stale numbers that look plausible.

4. 4

   Treat the warehouse copy as the joinable one, not the canonical one

   The dashboard remains the reference for ScaleBun's own metrics; a difference is usually a transformation in your pipeline.

Permissions and prerequisites#

Requires an administrative role and credentials for your warehouse with permission to write to the target.

Limits and edge cases#

• Delivery is at-least-once. Expect to deduplicate on your side.

• Late-arriving events land late. Offline clients flush when they reconnect, so a day's data is not final at midnight. Build your models to tolerate that.

• Schema evolves. New fields appear as capabilities ship. A pipeline that breaks on unknown columns will break eventually.

• The dashboard and your warehouse will disagree slightly at any given moment. That is delivery lag, not an error.

Troubleshooting#

Pipeline shows failed. Credentials or permissions at the destination, most often. Check that the target still exists and the account can write to it.

Row counts do not match the dashboard. Check for duplicates from at-least-once delivery, then for late-arriving events on the boundary day.

A new field is missing. The schema grew after your pipeline was modelled. Re-check the destination table definition.

Where the data comes from#

Served by

• Warehouse export

Related#

• Data export

• Data privacy

• Alerts & routing

• Webhooks

• Settings overview