Spark can be deployed in various modes. The most important ones are:
- Standalone mode: As an independent cluster not dependent upon any external cluster manager
- Amazon EC2: On EC2 instances of Amazon Web Services where it can access data from S3
- Apache YARN: The Hadoop ResourceManager
Other options include Apache Mesos and Kubernetes.
Further details can be found at the Spark documentation website, https://spark.apache.org/docs/latest/index.html.