SQL Warehouse Governance

SQL Warehouses That Don't
Burn Budget Overnight

Warehouses left running, oversized for the workload, with auto-stop disabled. Kostavo scans for all of it and notifies your team or stops them automatically.

Where Warehouse Costs Leak

Warehouses Left Running

A warehouse with no auto-stop keeps consuming compute even when no one is querying. Databricks doesn't enforce this for you.

Oversized for the Workload

Analysts spin up 2XL warehouses for exploratory queries. Without size policies, overprovisioning goes unnoticed.

Misconfigured & Exposed

Warehouses on the preview channel, with open permissions, missing cost tags, or excessive multi-cluster scaling create hidden risk.

What Kostavo Scans

Running & Idle Detection

Flags warehouses that are currently running and verifies whether auto-stop is disabled or set above your configured threshold. Can stop them automatically.

  • Running warehouse detection
  • Auto-stop disabled/too-high check
  • Configurable thresholds

Size & Scaling Policies

Scans warehouse size against your configured maximum and flags warehouses where the multi-cluster max exceeds a reasonable threshold.

  • Max warehouse size enforcement
  • Excessive multi-cluster detection
  • Automatic downscale option

Config & Security Policies

Detects warehouses on the preview channel, with overly permissive access, missing cost tags, or not using serverless compute.

  • Preview channel detection
  • Open permissions check
  • Missing tags & serverless nudge

Warehouse findings alongside the rest of your estate: running warehouses, sizing, and serverless candidates.

Warehouse Policies at a Glance

Running Warehouses

Currently running: notify or stop

Auto-Stop

Disabled or above max threshold: fix or notify

Warehouse Size

Exceeds maximum: downscale or notify

Multi-Cluster Scaling

Max clusters too high: fix or notify

Permissions & Channel

Open access or preview channel: notify

Tags & Serverless

Missing cost tags or not using serverless: notify

Alerts & Queries That Waste Warehouse Compute

Alerts on Oversized Warehouses

SQL Alerts (v2) running on warehouses larger than needed: can move to a smaller warehouse automatically

High-Frequency Alerts

SQL Alerts (v2) scheduled to run too frequently (wasteful polling against a warehouse): fix or notify

Alerts Without Destinations

SQL Alerts (v2) with no notification destination configured: silently running but alerting no one

Legacy SQL Alerts

Alerts using the deprecated legacy SQL alert format: flag for migration or removal

Stop Paying for Idle Warehouses

Connect a workspace and see which warehouses are sitting idle right now.

Start Free Trial