# Snowflake Virtual Warehouse

> The compute in Snowflake that runs your queries. You pay per second while it runs.

**Field:** Data Engineering  
**Category:** Storage & Warehousing  
**Also known as:** Virtual Warehouse, Snowflake Warehouse

## Think of it like

It is like a team of cooks you hire by the minute. A bigger team cooks faster but costs more.

## Definition

A virtual warehouse is a cluster of compute that Snowflake runs for your queries and loads. It holds no data, because the data sits in shared storage. Sizes run from X-Small to 6X-Large. X-Small uses one credit an hour, and each size up uses twice as many. Billing is per second, with a 60-second minimum each time it starts. By default it stops after 10 idle minutes and wakes when a query comes.

## Example

A team gives its dashboards an X-Small warehouse and its nightly load a Large one. The two never slow each other down.

## Example code

```sql
CREATE WAREHOUSE reports_wh
  WAREHOUSE_SIZE = XSMALL
  AUTO_SUSPEND = 60
  AUTO_RESUME = TRUE;
```

A small warehouse that stops after 60 idle seconds and wakes on the next query.

## Why it matters

Warehouses are most of a Snowflake bill. The size and the idle time decide what you pay.

## Common mistake

Making one big warehouse for everyone. Give the nightly load its own, so it cannot slow the dashboards.

## Three principles

- **Compute only.** The data lives in shared storage.
- **Doubling sizes.** Each size up, twice the credits.
- **Auto suspend.** Stops when idle, wakes on demand.

## Related terms

- [Snowflake](https://learn.benibix.com/term/data/snowflake)
- [Separation of Storage and Compute](https://learn.benibix.com/term/data/separation-of-storage-and-compute)
- [Data Warehouse](https://learn.benibix.com/term/data/data-warehouse)
- [Massively Parallel Processing (MPP)](https://learn.benibix.com/term/data/massively-parallel-processing)
- [Zero-Copy Clone](https://learn.benibix.com/term/data/zero-copy-clone)
- [Snowpark](https://learn.benibix.com/term/data/snowpark)

![Left, tied together: two servers, each with its own compute and its own disk, so more data means more servers even when nobody is querying. Right, kept apart: three compute boxes, one for reports, one for loading and one turned off at no cost, all reading one shared store of data. A note says you add compute for a busy hour and turn it off after, and the data never moves.](https://learn.benibix.com/diagrams/dat-storage-compute.svg)

---

Source: https://learn.benibix.com/term/data/snowflake-virtual-warehouse
Author: Behnia Zarei (a developer in Vienna, Austria, who builds and teaches data, AI and business intelligence at BENIBIX)
Publisher: BENIBIX Learn (https://learn.benibix.com/)
Last updated: 2026-11-16
Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Quote freely with attribution.
