Skip to content

Parallel Execution

Understanding parallelism, performance tuning, and best practices for batch operations.

How Parallelism Works

azlin uses thread pools to execute operations on multiple VMs simultaneously:

  • Default: 10 parallel workers
  • Queue remaining VMs
  • Continue on individual failures

Tuning Guidelines

Operation Type

Light (status, simple commands): 10-20 workers 

azlin batch command 'uptime' --all --max-workers 20

Medium (updates, file operations): 5-10 workers 

azlin batch sync --all --max-workers 8

Heavy (builds, migrations): 1-5 workers 

azlin batch command 'docker build .' --all --max-workers 3

Network Considerations

  • Azure Bastion connections: Reduce to 5-10 workers
  • Cross-region VMs: Reduce parallelism
  • Local region VMs: Can use higher parallelism

Performance Tips

  1. Start conservative: Begin with default (10), adjust based on results
  2. Monitor connections: Watch for SSH timeouts with high parallelism
  3. Consider timeouts: Long operations need higher timeouts
  4. Test first: Run on subset before full fleet

Example Tuning

# Too aggressive - may timeout
azlin batch command 'apt upgrade -y' --all --max-workers 20

# Better - balanced for heavy operation
azlin batch command 'apt upgrade -y' --all --max-workers 5 --timeout 900

See Also

Documentation last updated: 2025-11-24