Monolithic to Microservices @ Moip

Presented @ API Experience www.apiexperience.com.br

Transcript

1. MONOLITHIC TO MICROSERVICES
3. Um pouco sobre os nossos últimos anos: + 300 parceiros integrados + 100k clientes + R$ 1 bi / ano + 200 colaboradoradores + 8 anos de operação
6. V1
7. V2
8. MONOLITHIC VS MICROSERVICES
9. MOIP V1
12. LEGACY CODE
14. DIFFERENT TEAMS DEPLOY SYNCHRONIZATION
15. MOIP V2 FIRST PHASE
16. ‣ REST APIS USING V1 CODEBASE ‣ NEW FACADE TO LEGACY CODE ‣ RE-USE V1 FEATURES ‣ FAST TO DELIVER
19. HARD TO MAINTAIN HARD TO HANDLE ERRORS
20. 500 ERRORS
21. NOT FAST TO DELIVER
22. SLOW RESPONSE TIMES
23. MOIP V2 SECOND PHASE
24. ‣ SAME DATABASE ‣ NEW CODE ‣ UNIT / INTEGRATION TESTS ‣ EASIER TO MAINTAIN ‣ BETTER RESPONSE TIMES
27. PERSISTING COLUMNS THAT WE HAVE NO IDEA ABOUT THE MEANING
28. V1 DATA <——> V2 DATA
29. MOIP V2 THIRD PHASE
30. ‣ DATABASE FOR EACH SERVICE ‣ NEW CODE ‣ UNIT / INTEGRATION TESTS ‣ SYNC V2 <-> V1
35. SYNC DATA BETWEEN SERVICES IS VERY HARD
36. CONSISTENCY
37. MONITORING
38. NO DEVOPS
39. BUT…
40. ‣ SCALABILITY ‣ FASTER DEPLOYS ‣ RESILIENCY ‣ EASIER TO MAINTAIN
41. COMMUNICATION
42. SYNCHRONOUS VS ASYNCHRONOUS
43. REST API AMQP
44. SYNCHRONOUS ‣ LATENCY ‣ TIMEOUT ‣ CASCADING ERRORS ‣ SERVICE DOWN
45. NETFLIX STACK
46. EVENT DRIVEN ARCHITECTURE ‣ SERVICE NOTIFY EVENTS ‣ SERVICE SUBSCRIBE TO EVENTS ‣ HEADERS / JSON BODY
50. CONSISTENCY
51. DISTRIBUTED TRANSACTIONS WITHOUT TRANSACTIONS
53. HARD TO CHECK OVER DIFFERENT DATABASES
55. INFRASTRUCTURE
56. ‣ DEPLOY ‣ INFRASTRUCTURE AS CODE ‣ DEVELOPER DEPLOY TO PROD ‣ PLATFORM ‣ SERVICE DISCOVERY
57. MONITORING
58. INFRASTRUCTURE MONITORING ‣ AUTOMATION ‣ DASHBOARDS ‣ SERVER LOG AGGREGATION
59. BUSINESS MONITORING ‣ BUSINESS METRICS ‣ DASHBOARDS ‣ APPLICATION LOG AGGREGATION
60. LESSONS LEARNED
61. ‣ MICROSERVICE ONLY IF YOU NEED IT ‣ MICROSERVICE NOT NANOSERVICE ‣ AUTOMATE INFRASTRUCTURE ‣ METRICS, METRICS AND METRICS
62. ‣ TEST FAILURE SCENARIOS ‣ INTERNAL LATENCY ‣ USE SERVICE DISCOVERY
63. MOIP V2 FUTURE
64. ‣ CONNECTION CACHING ‣ CACHE HEADERS ‣ DOCKER CONTAINERS ‣ KEY/VALUE STORE CONFIGURATION ‣ SERVICE DISCOVERY ‣ FACEBOOK PRESTO
65. THANK YOU !!! Rodrigo Saito @rodrigosaito