Python in the MicroServices world

Transcript

1. Python in the MicroServices world @harikrishnan83
2. About me Full Stack Dev Tech Consulting for Startups Java, .Net, JavaScript, Python, Ruby Concurrency, High Performance Applications, REST @harikrishnan83
3. Why this talk?
4. “Microservices are a more concrete and modern interpretation of service-oriented architectures (SOA)” - Wikipedia MicroServices by Definition UI Core BI DB Admin and Dashboard API DB Booking Service API DB Payment Service API DB Notification Service API DB API DB API DB Message Bus UI
5. Why all the Buzz?
6. Small, Composable, Maintainable, Independently deployable etc.
7. Have we not seen this before?
8. JEE EJB (Message Driven Beans) MSG Queue MSG Java EE Server MDB Instances EJB Container Application Client
9. OSGi
10. Why is MicroServices architecture any different?
11. Do not put all your eggs in one basket
12. How does Python Stack up?
15. All Python MicroServices App?
16. Before we start
17. Your First MicroService
18. Monolithic Application Search by Country and State
19. Monolithic Application Search by Country and State API DB View Layer
20. Monolithic Application Search by Country and State API DB View Layer Key Value Store
21. Okay, there we have a MicroService. Now let us build some serious Features on the Monolithic App.
22. Monolithic Application Search by Country and State API DB View Layer Key Value Store New Feature API
23. A typical MicroService 1. API 2. Logic 3. Data 4. Dependencies 5. Automation 6. Admin 7. Monitoring 8. ...
24. Discipline 1. Iterative Development 2. Continuous Integration and Delivery 3. Release Pipelining and Build Promotion
25. Motivations to slice
26. Technical 1. Additional Infrastructure 2. Increasing Resilience 3. External Facing vs Internal Facing APIs
27. Business 1. A product by itself 2. Need to monitor the ROI of the feature in isolation 3. A/B testing 4. Build vs Buy
28. Organisational 1. Conway’s law organizations which design systems ... are constrained to produce designs which are copies of the communication structures of these organizations — M. Conway 2. Team size
29. Distributed Problems
30. Identity
31. Booking Service Payment Service Notification Service View User Service Auth And Auth Book Ticket User Details Send SMS Process Payment Anonymous Access Anonymous Access Back Office Dashboard
32. Service Traceability
33. Booking Service Payment Service Notification Service View User Service Auth And Auth Book Ticket Booking Ref No Send SMS Process Payment Booking Ref No Booking Ref No
34. Reconciliation
35. Distributed Transactions 1. Rollbacks are expensive and quite hard 2. Eventual Consistency
36. User Signup Wizard User Service Queue Failed Messages
37. Project Management
38. What happens to Sprints? 1 To Do Doing Done To Do Doing Done 1 To Do Doing Done 1 Payment Service Booking Service Notification Service Feature: Book Ticket
39. API
40. Sealing the Contract We had to get the stories to go in Parallel So we agreed on the API boundaries and started coding to Apiary However it did not quite work out Many details became clear only after we started building the feature
41. What do we do? 1. Follow the data 2. Understand Non Functional Aspects 3. Every time we add an API call, think if it belongs there
42. API 1. Synchronous - Simple Request Response 2. Asynchronous - Fire and Forget - Queues 3. Callbacks - Webhooks
43. Branch and Merge Strategy 1. Single Repository was not working for us a. Initially it was great because we just did feature branches b. Most of us make several small commits 2. Internal Open Source a. Service Owner b. Feature Owner c. Business Owner
44. Thank you @harikrishnan83