Multicluster warehouse

As explained in Snowflake Data Engineering

A multicluster warehouse is a Snowflake virtual warehouse configured to add or deactivate compute clusters according to specified parameters and demand.

A multicluster warehouse is a Snowflake virtual warehouse that adds compute clusters when demand increases and deactivates them when they are no longer needed, according to configured parameters. section 9.1.3

Use multicluster warehouses to reduce query queuing when many users or applications share a warehouse. Snowflake provisions additional clusters as workload demand requires and suspends them afterward. section 9.4.1 section 9.6

Choose the operational response according to workload behavior:

• Convert a single-cluster warehouse to a multicluster warehouse when concurrent workloads queue. section 9.4.1
• Increase the maximum cluster count when an existing multicluster warehouse still queues queries. section 9.4.1
• Accept queuing during noncritical nightly batch loads when it indicates that the warehouse is fully utilized. section 9.4.1

Inspect recent warehouse queuing

Query SNOWFLAKE.ACCOUNT_USAGE.WAREHOUSE_LOAD_HISTORY for the past seven days, grouping by date and warehouse to compare total running time with total queued load before changing cluster settings. section 9.4.1

Figure 9.2 Snowflake virtual warehouse sizes and their credit consumption per hour

Compare Snowflake warehouse sizes with their hourly credit consumption when evaluating the cost and capacity implications of additional compute. section 9.1.3

How Snowflake Data Engineering covers it

How deep it goes

Explains dynamic cluster activation, deactivation, queuing diagnosis, and operational choices without providing warehouse-creation syntax.

Examples

Uses WAREHOUSE_LOAD_HISTORY queries and warehouse-size credit comparisons to assess queuing and capacity.

What it assumes

Assumes knowledge of Snowflake virtual warehouses, concurrent queries, and querying usage views in the SNOWFLAKE database.

What you can do after

Diagnose query queuing with WAREHOUSE_LOAD_HISTORY and choose between added warehouses, multicluster conversion, or more clusters.

How it connects

Connects virtual warehouses, query queuing, compute costs, and WAREHOUSE_LOAD_HISTORY.

liveBookmanning.com

Read Snowflake Data Engineering on liveBook Get the book at manning.com

Related concepts

• clusteringuses · Clustering is the technique of grouping data items according to similarity or shared characteristics.
• queuingaddresses · Queuing is the practice of buffering and ordering work items or requests for processing when capacity becomes available or according to scheduling rules.

This explainer was generated by Manning's knowledge graph from the text of Snowflake Data Engineering. The section links point to the parts of the book it draws on.