staRburst

install.packages("starburst")

remotes::install_github("scttfrdmn/starburst")

library(starburst)

# One-time setup (2 minutes)
starburst_setup()

# Run parallel computation on AWS
results <- starburst_map(
  1:1000,
  function(x) expensive_computation(x),
  workers = 50
)
#> 🚀 Starting starburst cluster with 50 workers
#> 💰 Estimated cost: ~$2.80/hour
#> 📊 Processing 1000 items with 50 workers
#> 📦 Created 50 chunks (avg 20 items per chunk)
#> 🚀 Submitting tasks...
#> ✓ Submitted 50 tasks
#> ⏳ Progress: 50/50 tasks (3.2 minutes elapsed)
#>
#> ✓ Completed in 3.2 minutes
#> 💰 Estimated cost: $0.15

library(starburst)

# Define simulation
simulate_portfolio <- function(seed) {
  set.seed(seed)
  returns <- rnorm(252, mean = 0.0003, sd = 0.02)
  prices <- cumprod(1 + returns)

  list(
    final_value = prices[252],
    sharpe_ratio = mean(returns) / sd(returns) * sqrt(252)
  )
}

# Run 10,000 simulations on 100 AWS workers
results <- starburst_map(
  1:10000,
  simulate_portfolio,
  workers = 100
)
#> 🚀 Starting starburst cluster with 100 workers
#> 💰 Estimated cost: ~$5.60/hour
#> 📊 Processing 10000 items with 100 workers
#> ⏳ Progress: 100/100 tasks (3.1 minutes elapsed)
#>
#> ✓ Completed in 3.1 minutes
#> 💰 Estimated cost: $0.29

# Extract results
final_values <- sapply(results, function(x) x$final_value)
sharpe_ratios <- sapply(results, function(x) x$sharpe_ratio)

# Summary
mean(final_values)    # Average portfolio outcome
quantile(final_values, c(0.05, 0.95))  # Risk range

# Comparison:
# Local (single core): ~4 hours
# Cloud (100 workers): 3 minutes, $0.29

# Create cluster once
cluster <- starburst_cluster(workers = 50, cpu = 4, memory = "8GB")

# Run multiple analyses
results1 <- cluster$map(dataset1, analysis_function)
results2 <- cluster$map(dataset2, processing_function)
results3 <- cluster$map(dataset3, modeling_function)

# All use the same Docker image and configuration

# For memory-intensive workloads
results <- starburst_map(
  large_datasets,
  memory_intensive_function,
  workers = 20,
  cpu = 8,
  memory = "16GB"
)

# For CPU-intensive workloads
results <- starburst_map(
  cpu_tasks,
  cpu_intensive_function,
  workers = 50,
  cpu = 4,
  memory = "8GB"
)

# Start detached session
session <- starburst_session(workers = 50, detached = TRUE)

# Submit work and get session ID
session$submit(quote({
  results <- starburst_map(huge_dataset, expensive_function)
  saveRDS(results, "results.rds")
}))
session_id <- session$session_id

# Disconnect - job continues running
# Later (hours/days), reconnect:
session <- starburst_session_attach(session_id)
status <- session$status()  # Check progress
results <- session$collect()  # Get results

# Cleanup when done
session$cleanup(force = TRUE)

# Set cost limits
starburst_config(
  max_cost_per_job = 10,      # Hard limit
  cost_alert_threshold = 5     # Warning at $5
)

# Costs shown transparently
results <- starburst_map(data, fn, workers = 100)
#> 💰 Estimated cost: ~$3.50/hour
#> ✓ Completed in 23 minutes
#> 💰 Estimated cost: $1.34

results <- starburst_map(data, fn, workers = 100, cpu = 4)
#> ⚠ Requested 100 workers (400 vCPUs) but quota allows 25 workers (100 vCPUs)
#> ⚠ Using 25 workers instead
#> 💰 Estimated cost: ~$1.40/hour

starburst_config(
  region = "us-east-1",
  max_cost_per_job = 10,
  cost_alert_threshold = 5
)

@software{starburst,
  title = {staRburst: Seamless AWS Cloud Bursting for R},
  author = {Scott Friedman},
  year = {2026},
  version = {0.3.6},
  url = {https://starburst.ing},
  license = {Apache-2.0}
}