# Interaction of Cluster Components

You run database queries against an aggregator. Most of the communication between aggregators and leaves for query execution is also implemented as SQL statements.

![](https://cdn.sanity.io/images/h3bp8c98/production/879ff2450e79b107e72ae63a8cc4f307fe7383f5-1714x580.png)

## Parallel Execution of Queries

By default, SingleStore will create one partition per CPU core on the leaves for maximum parallelism. This number is configurable cluster-wide with the `default-partitions-per-leaf` variable, or as an optional parameter to [CREATE DATABASE](https://docs.singlestore.com/db/v9.0/reference/sql-reference/data-definition-language-ddl/create-database.md).

In the context of query execution, a partition is the granular unit of query parallelism. In other words, every parallel query is run with a level of parallelism equal to the number of partitions. You can view the partitions in a database with [SHOW PARTITIONS](https://docs.singlestore.com/db/v9.0/reference/sql-reference/show-commands/show-partitions.md).

***

Modified at: June 22, 2022

Source: [/db/v9.0/introduction/distributed-architecture/interaction-of-cluster-components/](https://docs.singlestore.com/db/v9.0/introduction/distributed-architecture/interaction-of-cluster-components/)

(An index of the documentation is available at /llms.txt)