Multi-cluster Warehouses

Multi-cluster warehouses enable you to scale compute resources to manage your user and query concurrency needs as they change, such as during peak and off hours.

In this Topic:

What is a Multi-cluster Warehouse?

By default, a virtual warehouse consist of a single cluster of servers that determines the total resources available to the warehouse for executing queries. As queries are submitted to a warehouse, the warehouse allocates resources to each query and begins executing the queries. If sufficient resources are not available to execute all the queries submitted to the warehouse, Snowflake queues the additional queries until the necessary resources become available.

With multi-cluster warehouses, Snowflake supports allocating, either statically or dynamically, a larger pool of resources to each warehouse. A multi-cluster warehouse is defined by specifying the following properties:

  • Maximum number of server clusters, greater than 1 (up to 10).
  • Minimum number of server clusters, equal to or less than the maximum (up to 10).

Additionally, multi-cluster warehouses support all the same properties and actions as single-cluster warehouses, including:

  • Specifying a warehouse size.
  • Resizing a warehouse at any time.
  • Auto-suspending a running warehouse due to inactivity; note that this does not apply to individual clusters, but rather the entire warehouse.
  • Auto-resuming a suspended warehouse when new queries are submitted.

Maximized vs. Auto-scale

You can choose to run a multi-cluster warehouse in either of the following modes:

Maximized:

This mode is enabled by specifying the same value for both maximum and minimum clusters (note that the specified value must be larger than 1). In this mode, when the warehouse is started, Snowflake starts all the clusters so that maximum resources are available while the warehouse is running.

This mode is effective for statically controlling the available resources (i.e. servers), particularly if you have large numbers of concurrent user sessions and/or queries and the numbers do not fluctuate significantly.

Auto-scale:

This mode is enabled by specifying different values for maximum and minimum clusters. In this mode, Snowflake starts and stops clusters as needed to dynamically manage the load on the warehouse:

  • As the number of concurrent user sessions and/or queries for the warehouse increases, and queries start to queue due to insufficient resources, Snowflake automatically starts additional clusters, up to the maximum number defined for the warehouse.
  • Similarly, as the load on the warehouse decreases, Snowflake automatically shuts down clusters to reduce the number of running servers and, correspondingly, the number of credits used by the warehouse.

To help control the usage of credits in Auto-scale mode, Snowflake provides a property, SCALING_POLICY, that determines the scaling policy to use when automatically starting or shutting down additional clusters. For more information, see Setting the Scaling Policy for a Multi-cluster Warehouse (in this topic).

Tip

When determining the maximum and minimum clusters to use for a warehouse, start with Auto-scale mode and start small (e.g. maximum = 2 or 3, minimum = 1). As you track how your warehouse load fluctuates over time, you can increase the maximum and minimum clusters until you determine the numbers that best support the upper and lower boundaries of your user/query concurrency.

Multi-cluster Size and Credit Usage

The number of servers in each cluster is determined by warehouse size:

  • The total number of servers for the warehouse is calculated by multiplying the warehouse size by the maximum number of clusters. This also indicates the maximum number of credits consumed by the warehouse per full hour of usage (i.e. if all clusters run during the hour).

    For example, the maximum number of credits consumed per hour for a Medium-size warehouse (4 servers per cluster) with 3 clusters is 12 credits.

  • If a multi-cluster warehouse is resized, the new size applies to all the clusters for the warehouse, including clusters that are currently running and any clusters that are started after the warehouse is resized.

The actual number of credits consumed per hour depends on the number of clusters running during each hour that the warehouse is running. For more details, see Examples of Multi-cluster Credit Usage (in this topic).

Benefits of Multi-cluster Warehouses

With a standard, single-cluster warehouse, if your user/query load increases to the point where you need more compute resources:

  1. You must either increase the size of the warehouse or start additional warehouses and explicitly redirect the additional users/queries to these warehouses.
  2. Then, when the resources are no longer needed, to conserve credits, you must manually downsize the larger warehouse or suspend the additional warehouses.

In contrast, a multi-cluster warehouse enables larger numbers of users to connect to the same size warehouse. In addition:

  • In Auto-scale mode, a multi-cluster warehouse eliminates the need for resizing the warehouse or starting and stopping additional warehouses to handle fluctuating workloads. Snowflake automatically starts and stops additional clusters as needed.
  • In Maximized mode, you can control the capacity of the warehouse by increasing or decreasing the number of clusters as needed.

Tip

Multi-cluster warehouses are best utilized for scaling resources to improve concurrency for users/queries. They are not as beneficial for improving the performance of slow-running queries or data loading. For these types of operations, resizing the warehouse provides more benefits.

Examples of Multi-cluster Credit Usage

The following four examples illustrate credit usage for a multi-cluster warehouse.

Note

For the sake of simplicity, all these examples depict credit usage in increments of 1 hour, 30 minutes, and 15 minutes. In a real-world scenario, with per-second billing, the actual credit usage would contain fractional amounts, based on the number of seconds that each cluster of servers runs.

Example 1: Maximized (2 Hours)

In this example, a Medium-size warehouse (4 servers per cluster) with 3 clusters runs in Maximized mode for 2 hours:

  Cluster 1 Cluster 2 Cluster 3 Total Credits
1st Hour 4 4 4 12
2nd Hour 4 4 4 12
Total Credits 8 8 8 24

Example 2: Auto-scale (2 Hours)

In this example, a Medium-size warehouse (4 servers per cluster) with 3 clusters runs in Auto-scale mode for 2 hours:

  • Cluster 1 runs continuously.
  • Cluster 2 runs continuously for the 2nd hour only.
  • Cluster 3 runs for 30 minutes during the 2nd hour.
  Cluster 1 Cluster 2 Cluster 3 Total Credits
1st Hour 4 0 0 4
2nd Hour 4 4 2 10
Total Credits 8 4 2 14

Example 3: Auto-scale (3 Hours)

In this example, a Medium-size warehouse (4 servers per cluster) with 3 clusters runs in Auto-scale mode for 3 hours:

  • Cluster 1 runs continuously.
  • Cluster 2 runs continuously for the entire 2nd hour and 30 minutes in the 3rd hour.
  • Cluster 3 runs for 30 minutes in the 3rd hour.
  Cluster 1 Cluster 2 Cluster 3 Total Credits
1st Hour 4 0 0 4
2nd Hour 4 4 0 8
3rd Hour 4 2 2 8
Total Credits 12 6 2 20

Example 4: Auto-scale (3 Hours) with Resize

In this example, the same warehouse from example 3 runs in Auto-scale mode for 3 hours with a resize from Medium (4 servers per cluster) to Large (8 servers per cluster):

  • Cluster 1 runs continuously.
  • Cluster 2 runs continuously for the 2nd and 3rd hours.
  • Warehouse is resized from Medium to Large at 2:30 hours.
  • Cluster 3 runs for 15 minutes in the 3rd hour.
  Cluster 1 Cluster 2 Cluster 3 Total Credits
1st Hour 4 0 0 4
2nd Hour 4+2 4+2 0 12
3rd Hour 8 8 2 18
Total Credits 18 14 2 34

Creating a Multi-Cluster Warehouse

You can create a multi-cluster warehouse through the web interface or using SQL:

Interface:

Click on Warehouses » Create:

  1. In the Maximum Clusters field, select a value greater than 1.
  2. In the Minimum Clusters field, optionally select a value greater than 1.
  3. Enter other information for the warehouse, as needed, and click Finish.
SQL:

CREATE WAREHOUSE with:

  • MAX_CLUSTER_COUNT set to a value greater than 1.
  • MIN_CLUSTER_COUNT (optionally) set to a value greater than 1.

To view information about the multi-cluster warehouses you create:

Interface:

Click on Warehouses

The Clusters column displays the minimum and maximum clusters for each warehouse, as well as the number of clusters that are currently running if the warehouse is started.

SQL:

SHOW WAREHOUSES

The output includes three columns (min_cluster_count, max_cluster_count, started_clusters_column) that display the same information provided in the Clusters column in the web interface.

All other tasks for multi-cluster warehouses (except for the remaining tasks described in this topic) are identical to single-cluster warehouse tasks.

Setting the Scaling Policy for a Multi-cluster Warehouse

To help control the credits consumed by a multi-cluster warehouse running in Auto-scale mode, Snowflake provides scaling policies, which are used to determine when to start or shut down a cluster.

The scaling policy for a multi-cluster warehouse only applies if it is running in Auto-scale mode. In Maximized mode, all clusters run concurrently so there is no need to start or shut down individual clusters.

Snowflake supports the following scaling policies:

Policy Description Cluster Starts… Cluster Shuts Down…
Standard (default) Prevents/minimizes queuing by favoring starting additional clusters over conserving credits. Immediately when either a query is queued or the system detects that there’s one more query than the currently-running clusters can execute. After 2 to 3 consecutive successful checks (performed at 1 minute intervals), which determine whether the load on the least-loaded cluster could be redistributed to the other clusters without spinning up the cluster again.
Economy Conserves credits by favoring keeping running clusters fully-loaded rather than starting additional clusters, which may result in queries being queued and taking longer to complete. Only if the system estimates there’s enough query load to keep the cluster busy for at least 6 minutes. After 5 to 6 consecutive successful checks (performed at 1 minute intervals), which determine whether the load on the least-loaded cluster could be redistributed to the other clusters without spinning up the cluster again.

Note

A third scaling policy, Legacy, was provided for backward compatibility. In contrast to the other policies, it used a static approach based on length of time a cluster is active/inactive.

Legacy has been obsoleted/removed. All warehouses that were using the Legacy policy now use the default Standard policy.

The scaling policy for a multi-cluster warehouse can be set when the warehouse is created or at any time afterwards, either through the web interface or using SQL:

Interface:

Click on:

  • Warehouses » Create or
  • Warehouses » warehouse_name » Configure

In the Scaling Policy field, select the desired value from the drop-down list.

SQL:

CREATE WAREHOUSE or ALTER WAREHOUSE with SCALING_POLICY set to the desired value.

For example, in SQL:

ALTER WAREHOUSE mywh SET SCALING_TYPE = 'ECONOMY';

Increasing or Decreasing Clusters for a Multi-cluster Warehouse

You can increase or decrease the number of clusters for a warehouse at any time, even while the warehouse is running and executing statements. Clusters can be increased or decreased for a warehouse through the web interface or using SQL:

Interface:Click on Warehouses » warehouse_name » Configure
SQL:ALTER WAREHOUSE

The effect of changing the maximum and minimum clusters for a running warehouse depends on whether it is running in Maximized or Auto-scale mode:

  • Maximized:
    ↑ max & min:Specified number of clusters start immediately.
    ↓ max & min:Specified number of clusters shut down when they finish executing statements and the auto-suspend period elapses.
  • Auto-scale:
    ↑ max:If new_max_clusters > running_clusters, no changes until additional clusters are needed.
    ↓ max:If new_max_clusters < running_clusters, excess clusters shut down when they finish executing statements and the scaling policy conditions are met.
    ↑ min:If new_min_clsuters > running_clusters, additional clusters immediately started to meet the minimum.
    ↓ min:If new_min_clusters < running_clusters, excess clusters shut down when they finish executing statements and the scaling policy conditions are met.

Monitoring Multi-cluster Warehouses

You can monitor usage of multi-cluster warehouses through the web interface:

Interface:

Click on History or Worksheet

These pages include a column, Cluster Number, that specifies the cluster used to execute the statements submitted to each warehouse.