Pick Your Battles

As presented on PolyConf 2016

Transcript

1. Pick Your Battles Zef Hemel (@zef)
3. February 5th 2012
4. February 5th 2012 2 a.m. CET
6. “All systems down!”
13. 0.4
14. require("cluster")
15. Master Worker Worker Worker Worker load distribution even http requests
16. Master Worker Worker Worker Worker load distribution 94% 2% 2% 2% CPU http requests
17. 2/4 node.js core devs (later 3/4)
20. 0.4 custom
22. Cool story bro’
24. Why do start-ups exist?
25. 1. Find product market-fit Why do start-ups exist?
26. 1. Find product market-fit 2. Move fast Why do start-ups exist?
27. 1. Find product market-fit 2. Move fast 3. Try out bleeding edge tech Why do start-ups exist?
29. Cycle.js AWS Lambda Rust
30. 1. Find product market-fit 2. Move fast 3. Try out bleeding edge tech Why do start-ups exist?
31. 1. Find product market-fit 2. Move fast 3. Try out bleeding edge tech Why do start-ups exist?
32. Pick Your Battles
34. How to organize 200k+ lines of JavaScript?
35. API breakage between node.js versions
36. Releases without downtime?
37. Exceptions crashing the node.js server
38. Freeze-ups (node.js and Redis)
40. 1. Find product market-fit 2. Move fast 3. Try out bleeding edge tech Why do start-ups exist?
41. 1. Find product market-fit 2. Move fast 3. Try out bleeding edge tech Why do start-ups exist? Avoid distractions
42. Pick Your Battles
43. “Anything that can go wrong will go wrong.”
44. “Anything that can go wrong will go wrong.” — Murphy’s Law
45. Murphy never operated at scale
46. “Anything that can go wrong will immediately blow up in your face.” — Zef’s Law
47. 5 WHYs
48. Our server crashed
49. Our server crashed Why did it crash?
50. Our server crashed Why did it crash? We didn’t handle all error cases
51. Our server crashed Why did it crash? We didn’t handle all error cases Why did we not handle all error cases?
52. Our server crashed Why did it crash? We didn’t handle all error cases Why did we not handle all error cases? We only have happy-path tests
53. Our server crashed Why did it crash? We didn’t handle all error cases Why did we not handle all error cases? We only have happy-path tests Why do we only have happy-path tests?
54. Our server crashed Why did it crash? We didn’t handle all error cases Why did we not handle all error cases? We only have happy-path tests Why do we only have happy-path tests? We don’t have time to write more tests
55. Our server crashed Why did it crash? We didn’t handle all error cases Why did we not handle all error cases? We only have happy-path tests Why do we only have happy-path tests? We don’t have time to write more tests Why don’t we have more time to write tests?
56. 5 WHYs
57. 5 WHYs Root-cause analysis
58. 5 REALLYs
59. We should use node.js, because it’s cool, fast and scalable!
60. We should use node.js, because it’s cool, fast and scalable! Really?
61. We should use node.js, because it’s cool, fast and scalable! Really? Yes.
62. We should use node.js, because it’s cool, fast and scalable! Really? Yes. Really?
63. We should use node.js, because it’s cool, fast and scalable! Really? Yes. Really? Ehm, well, yes.
64. We should use node.js, because it’s cool, fast and scalable! Really? Yes. Really? Ehm, well, yes. Really?
65. We should use node.js, because it’s cool, fast and scalable! Really? Yes. Really? Yes, well, I think so… Ehm, well, yes. Really?
66. We should use node.js, because it’s cool, fast and scalable! Really? Yes. Really? Yes, well, I think so… Really? Ehm, well, yes. Really?
67. We should use node.js, because it’s cool, fast and scalable! Really? Yes. Really? Yes, well, I think so… Really? Well, speed is not that important, but scale is. Ehm, well, yes. Really?
68. We should use node.js, because it’s cool, fast and scalable! Really? Yes. Really? Yes, well, I think so… Really? Well, speed is not that important, but scale is. Ehm, well, yes. Really? Really?
69. We should use node.js, because it’s cool, fast and scalable! Really? Yes. Really? Yes, well, I think so… Really? Well, speed is not that important, but scale is. Ehm, well, yes. Really? Really? Well, no. We only expect to have about 200 users.
70. 5 REALLYs: How To
71. 5 REALLYs: How To Ask “Really?” until the person in question understands that he or she is wrong. (usually no more than 5 reallys required.)
72. 5 REALLYs Force thinking; question assumptions
73. Zef’s advice
74. most boring technology
75. years old
81. don’t underestimate the value of mature technology
82. Pick Your Battles
83. Really?
84. 1. Find product market-fit 2. Move fast Why do start-ups exist?
85. 1. Find product market-fit 2. Move fast Why do start-ups exist? bleeding edge
 tech
86. don’t optimize prematurely
87. DynamoDB
88. you cannot predict the future,
89. you cannot predict the future, so don’t even try
90. DynamoDB
91. DynamoDB
92. be pragmatic
93. TL;DL
94. Don’t take your tech choices too lightly
95. Pick Your Battles http://zef.me https://twitter.com/zef Join me @