HiveOps Docs
Database Service

Zero-ETL mirror

An analytical copy of a transactional database

This records intent today. Turning it on stores the decision, and nothing creates a mirror yet. The rest of this page describes what it is for and what turning it on will do.

A transactional database is built for the short reads and writes an application makes while serving users. An analytical query, the kind that scans a great deal of data to answer one question, competes with that work and slows it down.

The mirror exists so those queries never touch the database serving your users. It keeps a real time analytical copy in Data Lake, read only, maintained for you.

"Zero-ETL" means there is no pipeline for you to build or operate. You do not schedule an export, write a transform, or reconcile a copy that drifted.

Turning it on

It is a switch on the database page. There is nothing to configure, because the mirror mirrors the whole database.

The mirror is read only. Writes go to the transactional database, and the mirror follows.

Turning it off

The mirror disappears. It is a copy, so nothing you cannot rebuild is lost, but anything querying it stops finding it.

Deleting the transactional database takes its mirror with it.

When to use it

Turn it on when analytical queries are competing with your application. Reporting, dashboards, and questions across the whole history of a table are the usual reasons.

Leave it off if everything you run touches a few rows. A mirror you do not query is a copy nobody reads.

On this page