starburst_cluster: Create a Starburst Cluster

In starburst: Seamless AWS Cloud Bursting for Parallel R Workloads

Create a Starburst Cluster

Description

Creates a cluster object for managing AWS workers (EC2 by default, or Fargate) using the staRburst Future backend.

Usage

starburst_cluster(
  workers = 10,
  cpu = 4,
  memory = "8GB",
  region = NULL,
  timeout = 3600,
  launch_type = "EC2",
  instance_type = "c7g.xlarge",
  use_spot = TRUE
)

Arguments

workers	Number of parallel workers
cpu	CPU units per worker
memory	Memory per worker
region	AWS region
timeout	Maximum runtime in seconds
launch_type	Compute backend: "EC2" (default) or "FARGATE"
instance_type	EC2 instance type when launch_type = "EC2" (default: "c7g.xlarge"). Worker CPU architecture follows the instance type (Graviton *g.* = ARM64, Intel/AMD = x86_64); there is no separate platform argument.
use_spot	Use EC2 Spot instances for cost savings (default: TRUE)

Value

A starburst_cluster object

Examples

if (starburst_is_configured()) {
  cluster <- starburst_cluster(workers = 20)
  results <- cluster$map(data, function(x) x * 2)

  # Fargate backend instead of the EC2 default
  fg <- starburst_cluster(workers = 20, launch_type = "FARGATE")
}

starburst documentation built on July 23, 2026, 1:07 a.m.