Highly available, distributed & resilient software... team

Design patterns for building distributed, resilient, and scalable web applications are well known today. However, setting up a project organization to build them effectively remains a challenge, especially when the team itself is distributed. The "human" factor is a source of problems that are difficult to predict.

What would happen if you could take inspiration from the architecture of your applications to improve the communication of your teams?

Microservices, message queues, load balancers, caches, sharding, decoupling... These architectural concepts are designed to improve digital exchanges within our apps, and above all to anticipate and manage failures.
By modeling the exchanges within your team as software architecture, we will see how these design patterns can also be effective in solving our communication issues, very human this time.

Transcript

1. Highly available, Highly available, Highly available, distributed distributed distributed & resilient software... & resilient software... & resilient software... team team team | Yohan Lasorsa | @sinedied
2. Back story @sinedied
3. What this talk is (really) about What this talk is (really) about What this talk is (really) about @sinedied
4. Who am I? Yohan Lasorsa Web tech geek, OSS maintainer, DIY lover @sinedied [ Cloud Developer Advocate @ ]
5. My team
6. Agenda Geographic distribution Scalability Reliability @sinedied
7. Mind Map Data Information Service T eam @sinedied
8. Geographic distribution @sinedied
9. (Some) Problems LATENCY A ? @sinedied
10. (Some) Problems SYNCHRONIZATION A B @sinedied
11. (Some) Problems COHERENCY ??? A B @sinedied
12. Possible solutions @sinedied
13. Overcoming latency CDN / Cache @sinedied
14. Overcoming latency Aggregation ? ? ? ? without with @sinedied
15. Overcoming coherency Centralization @sinedied
16. Overcoming coherency @sinedied
17. Overcoming "synchronization" T ransaction Wait! ? Ok! Ok? 1. 2. 3. @sinedied
18. Translation to the "human" world ? @sinedied
19. CDN / Cache Aggregation Centralization T ransaction All hands, Recordings, Minutes nohello.com Emails w/ Conte xt Knowledgebase (wiki) Meetings PM, T eam Lead, Manager PM, T eam Lead, Manager Backlog @sinedied
20. Human problems @sinedied
21. https://everytimezone.com @sinedied
22. Synchronous communication is difficult @sinedied
23. Scalability @sinedied
24. (Some) Problems ? ? ? ? ? ? BOOM! Overload @sinedied
25. (Some) Problems Communication @sinedied
26. Possible solutions @sinedied
27. Overcoming load issues Vertical scaling Horizontal scaling @sinedied
28. Overcoming load issues Monolith Microservices @sinedied
29. Overcoming load issues Sharding @sinedied
30. Overcoming communication issues Decoupling ? ? ? ? @sinedied
31. Overcoming communication issues Map / reduce @sinedied
32. Translation to the "human" world ? @sinedied
33. Scaling Microservices Small teams Recruitment, New teams Sharding Map/Reduce Autonomy Decoupling T eam specialty Manager "Ask Me Anything", Minutes, Debriefs T eam-specific OKRs @sinedied
34. Reliability @sinedied
35. (Some) Problems High availability @sinedied
36. (Some) Problems Data loss BOOM! @sinedied
37. (Some) Problems Monitoring 49 1 @sinedied
38. Possible solutions @sinedied
39. Implementing high availability Load balancing Failover @sinedied
40. Overcoming data loss: storage Backup A A' @sinedied
41. Overcoming data loss: writing Queue Ok! @sinedied
42. Overcoming data loss: reading Retry ? ? Time @sinedied
43. Implementing monitoring Health checks I'm ok! I'm ok! I'm ok! A B C D A B C D @sinedied
44. Implementing monitoring Measures Metrics 459 30% Alerts 30% @sinedied
45. Translation to the "human" world ? @sinedied
46. Load balancing / failover Backup Health checks Teammates Teammates :) Queue Monitoring Retry Status reports, Pings, Surveys, eNPS, 1:1 Manager, Teammates OKRs, Shared dashboard, Transparency! Backlog, Todo list Time zones, Docs, Pair prog., Sharing Knowledgebase, Docs Handover emails Follow-up, Try different channels Leadership team @sinedied
47. There's more... @sinedied
48. Security Persistence Updating Openness Heterogeneity ... @sinedied
49. We're human (after all) @sinedied
50. We're human (after all) Woot! Woot! @sinedied
51. /@sinedied THANK YOU! Slides bit.ly/distributeam