Amazon Simple Workflow
Embed Size (px)
Amazon Simple Workflow (SWF) is a new cloud-based workflow management service from Amazon. SWF is designed to build scalable, resilient applications, model and automate business processes, manage infrastructure services and build analytics pipelines. In this session we introduce SWF and walk through some real world use cases and examples.
Transcript of Amazon Simple Workflow
- 1. AmazonSimple Workow Launched yesterday!
- 2. Build, process andmanage workows
- 3. Workows?
- 4. Managing infrastructureBusiness process Analytics pipelines Workows?
- 5. Managing infrastructureBusiness process Analytics pipelines Workows? Applications
- 6. Register billing Create details accountNew customer Register address Validate billing details
- 7. Synchronous Register billing Create details accountNew customer Register address Validate billing details Asynchronous
- 8. Billing Account registration creation Q1 Q2 Q3New customer Address Billing registration validation
- 9. Ordering anddependencies
- 10. Duplication of messages
- 11. Recording metadata
- 12. Undifferentiated heavy lifting
- 13. Enter SWF!
- 14. Designed for the cloud
- 15. Scalable, resilient applications
- 16. Complex business processes
- 17. Easy to use
- 18. Flexible
- 19. Task orchestration
- 20. ManagedBased on application logic
- 21. DependenciesTask execution Managed Based on application logic Concurrency Scheduling
- 22. Zero administrationNo hardware of software to admin for orchestration
- 23. Vocabulary Domain Workow Action Activity workers Decider
- 24. VocabularyCollection of workflows Domain Workow Action Activity workers Decider
- 25. Vocabulary Domain Workow Collection of actions Action Activity workers Decider
- 26. Vocabulary Domain WorkowTask or workflow step Action Activity workers Decider
- 27. Vocabulary Domain Workow Action Activity workers Implement actions Decider
- 28. Vocabulary Domain Workow Action Activity workersCoordinate workflow Decider actions
- 29. Clean separationDecoupled task logic from application ow
- 30. ScaleWorkers and deciders are stateless
- 31. Signals and errorsCongurable timeouts and out-of-band signalling. Markers for checkpointing.
- 32. PAYG
- 33. Pricing $0.0001 per workow $0.000005 per day for > 24 hours $0.000025 per task, signal or marker $0.10 for rst Gb inboundFirst Gb out is free, normal AWS usage after that
- 34. Free tier 1000 workow executions10,000 tasks, timers, signals and markers 30,000 workow-days
- 35. Flow Framework Abstraction for Java
- 36. Left image Right image Warp Warp Stereo match 3D map Tiling
- 37. Q&Amatthew@amazon.com