High Performance Architecture Patterns

Transcript

1. High Performance Architecture Patterns Celso Crivelaro @celsocrivelaro
2. $ whoami > Celso Crivelaro - Engineer and MSc in Computer Engineering - Dev Ruby / Python / Java @celsocrivelaro http://crivelaro.me
5. SMTP Locaweb - Transaction email platform - Fast and high quality email sending - Live reports - Integration by SMTP Protocol and HTTP API - 2,5 Millions of messages per day - + 2 years
7. Platforms
12. High Availability
13. Resource Consumption
15. Disk Memory Band- width Processor time Logs GC Open Connections Deadlocks Stop all disk activities Swap Slowness Services stop responding Queues Amount Workers Slowness
16. Integrations
17. Integrations always fails!
18. Integration Fails always Internal External DB API Storage Network API ISP Hop Cloud
19. Map hidden integrations E.g. DNS
20. Fail Fast
21. Failing Fast reduces transient errors
22. Timeouts
23. Never trust in default timeouts
24. Threads Timeouts to get rid of deadlocls Internal Little timeout External Depends: The least possible
25. Retry
26. 1 2 3 5 8 n Fibonacci Proportion ...
27. Isolate Critical Components
28. Panel / API Painel Web Internal API - HTTP Resources Connection Pool App Server Business Rules Integrations
29. App Internal API App Panel Painel Web API - HTTP Resources Conn. Pool App Server Business Rules Integrations Resources Conn. Pool App Server Business Rules Integrations
30. Less Components == Less Problems
32. Redundancy
33. Health Check
36. Scalability
37. CAP Theorem
38. Availability Consistency Tolerance to network partition
39. Availability Consistency Partition Tolerance to network partition whether system is available or not
40. Availability Consistency Tolerance to network partition Same data in all nodes
41. Availability Consistency Tolerance to network partition Tolerante when a network fails
42. Availability Consistency Partition Tolerance Postgres
43. Availability Consistency Partition Tolerance DNS
44. Availability Consistency Partition Tolerance Protocols
45. Evaluate which one is acceptable to your platform
46. Pool of Resources
47. App Connection Resource Connection Connection Connection Connection
48. Controlled Timeouts Controlled Resource Consumption Avoid opening/closing connections overhead
49. Cache
50. Make your data closer to your app
51. When use cache App API DB
52. Default Cache Strategy App Cache Source 1 2 3
53. Hot Cache Strategy App Cache Source 2 1
54. Load Balacing
55. LB App App App App
56. Load Balance Strategy matters!
57. Decoupled Architecture
58. Slow Processing Fast processing Asynchronous Processing Authentication Message Transformation MTA Providers User Message Persistence Consolidation
59. Background jobs Messaging frameworks
60. How to discover bottlenecks?
61. Test your limits... Load Test Stress Test Peak Test
62. Telemetry
64. Grafana + InfluxDB
65. Other scalabity patterns
66. Indexes
67. Sharding
68. References
72. http://use-the-index-luke.com
74. Thank you! Celso Crivelaro @celsocrivelaro celsocrivelaro@gmail.com