Add Lakebase Autoscaling (Postgres project) example

knowledge_base/postgres_autoscaling/README.md

@@ -0,0 +1,53 @@
+# Lakebase Autoscaling (Postgres) project with branching
+
+This example demonstrates how to define a Lakebase Autoscaling project with a non-default branch in a Declarative Automation Bundle.
+
+It includes and deploys:
+- A Lakebase Autoscaling project with configurable min and max compute units (CUs)
+- A non-default `development` branch for isolated dev/test workflows
+- A read-only replica endpoint on the default production branch
+
+Lakebase Autoscaling is Databricks' managed PostgreSQL service with autoscaling compute, Git-like branching, scale-to-zero, and instant point-in-time restore.
+
+For more information about Lakebase Autoscaling, see the [documentation](https://docs.databricks.com/aws/en/oltp/projects/).
+
+## Prerequisites
+
+* Databricks CLI v0.287.0 or above
+* `psql` client version 14 or above (Optional, only needed to run the demo queries)
+
+## Usage
+
+Modify `databricks.yml`:
+* Update the `host` field under each target's `workspace` block (`dev` and `prod`) to point at the Databricks workspace you want to deploy to
+* Adjust `autoscaling_limit_min_cu` and `autoscaling_limit_max_cu` to fit your workload (valid range: 0.5-32 CU, max minus min cannot exceed 16 CU)
+* Adjust `suspend_timeout_duration` to control scale-to-zero behavior (minimum: 60 seconds)
+
+The bundle defines two targets: `dev` (the default) and `prod`. Run `databricks bundle deploy` to deploy to `dev`, or `databricks bundle deploy --target prod` to deploy to `prod`.
+
+Please note that after this bundle gets deployed, the project and its compute endpoints start running, which incurs cost. Endpoints with scale-to-zero enabled will suspend after the configured `suspend_timeout_duration`.
+
+Run the following queries against the **production** branch to populate your database with sample data:
+
+```bash
+# Create a demo table:
+databricks psql --project my-autoscaling-project -- -d databricks_postgres -c "CREATE TABLE IF NOT EXISTS hello_world (id SERIAL PRIMARY KEY, message TEXT, number INTEGER, created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP);"
+
+# Insert 100 rows of demo data:
+databricks psql --project my-autoscaling-project -- -d databricks_postgres -c "INSERT INTO hello_world (message, number) SELECT 'Hello World #' || generate_series, generate_series FROM generate_series(1, 100);"
+
+# Show generated rows:
+databricks psql --project my-autoscaling-project -- -d databricks_postgres -c "SELECT * FROM hello_world;"
+```
+
+Run the following queries against the **development** branch to verify the branched data:
+
+**Important:** Prior to running the following command, the `development` branch MUST be 'reset' via the web UI. This will re-sync the child branch with its parent. More information regarding resetting a branch can be found [here](https://docs.databricks.com/aws/en/oltp/projects/manage-branches#reset-branch-from-parent).
+
+```bash
+# Query the development branch:
+databricks psql --project my-autoscaling-project --branch development -- -d databricks_postgres -c "SELECT count(*) FROM hello_world;"
+```
+
+## Clean up
+To remove the provisioned project, branches, and endpoints run `databricks bundle destroy`

knowledge_base/postgres_autoscaling/databricks.yml

@@ -0,0 +1,52 @@
+bundle:
+  name: postgres-autoscaling-example
+
+resources:
+  # Top-level project container for Lakebase Autoscaling.
+  # Creating a project automatically provisions:
+  #   - A default branch named "production"
+  #   - A primary READ_WRITE endpoint on the default branch
+  #   - A database with the name "databricks_postgres"
+  postgres_projects:
+    my_project:
+      project_id: my-autoscaling-project
+      display_name: "My Autoscaling Project"
+      pg_version: 17
+      default_endpoint_settings:
+        # Newly created endpoints inherit these settings unless overridden
+        #   at the endpoint level (see "prod_read_replica" below).
+        #   Changing these will not affect existing endpoints.
+        autoscaling_limit_min_cu: 1
+        autoscaling_limit_max_cu: 4
+        suspend_timeout_duration: "300s" # Scale-to-zero after 5 minutes
+
+  # (Optional) Read replica on the default (production) branch.
+  postgres_endpoints:
+    prod_read_replica:
+      parent: ${resources.postgres_projects.my_project.id}/branches/production
+      endpoint_id: ep-production-read-replica
+      endpoint_type: ENDPOINT_TYPE_READ_ONLY
+      # Note: The below two options override the ones set in
+      #   "default_endpoint_settings" above
+      autoscaling_limit_min_cu: 0.5
+      autoscaling_limit_max_cu: 2
+
+  # (Optional) Non-default branch for development/testing.
+  # Note: A READ_WRITE endpoint is currently created by default
+  #   when provisioning a new branch
+  postgres_branches:
+    dev:
+      parent: ${resources.postgres_projects.my_project.id}
+      branch_id: development  # The name for the branch
+      no_expiry: true
+
+# Targets allow you to deploy the same bundle to different Databricks workspaces.
+targets:
+  dev:
+    default: true
+    mode: development
+    workspace:
+      host: https://my-dev-workspace.cloud.databricks.com
+  prod:
+    workspace:
+      host: https://my-prod-workspace.cloud.databricks.com