Boston 2013 - Session - Aaron Quint

Transcript

1. CORRELATION: THE NEXT FRONTIER monitorama 2013 / boston / @aq Wednesday, May 29, 13
2. CTO of Wednesday, May 29, 13
3. Chief Taco Officer CTO of Wednesday, May 29, 13
4. Wednesday, May 29, 13
5. Wednesday, May 29, 13
6. A litte bit of @aq Wednesday, May 29, 13
7. A litte bit of @aq • Expert Eater Wednesday, May 29, 13
8. A litte bit of @aq • Expert Eater • Experienced Ruby and JS Developer Wednesday, May 29, 13
9. A litte bit of @aq • Expert Eater • Experienced Ruby and JS Developer • Growing Student of Operations Wednesday, May 29, 13
10. A litte bit of @aq • Expert Eater • Experienced Ruby and JS Developer • Growing Student of Operations • Beginner Distributed Systems Maintainer Wednesday, May 29, 13
11. Give each other dap. WE did it Wednesday, May 29, 13
12. OPS DONE. WE have DATA Y’ALL Wednesday, May 29, 13
13. Wednesday, May 29, 13
14. Uhhh SO NO MORE TEARS, right? Wednesday, May 29, 13
15. Wednesday, May 29, 13
16. THE LEVELS OF MONITORING NIRVANA Wednesday, May 29, 13
17. Wednesday, May 29, 13
18. PURE DATA Wednesday, May 29, 13
19. PURE DATA BASIC INFERENCES AND CORRELATIONS Wednesday, May 29, 13
20. PURE DATA BASIC INFERENCES AND CORRELATIONS THE FUCKING MATRIX Wednesday, May 29, 13
21. PURE DATA BASIC INFERENCES AND CORRELATIONS THE FUCKING MATRIX Wednesday, May 29, 13
22. PURE DATA BASIC INFERENCES AND CORRELATIONS THE FUCKING MATRIX whoa Wednesday, May 29, 13
23. Wednesday, May 29, 13
24. PURE DATA Wednesday, May 29, 13
25. PURE DATA BASIC INFERENCES AND CORRELATIONS Wednesday, May 29, 13
26. PURE DATA BASIC INFERENCES AND CORRELATIONS PREDICTIVE AND DIRECT RELATIONSHIPS Wednesday, May 29, 13
27. Aligning the data. CORRELATION Wednesday, May 29, 13
28. Except when it does. CORRELATION DOES NOT IMPLY CAUSATION Wednesday, May 29, 13
29. The marshmallow test Wednesday, May 29, 13
30. And let us get back to Shaving Yaks. correlation can narrow our work Wednesday, May 29, 13
31. Step back, HE’S DOING MATH! MATHEMATICAL CORRELATION Wednesday, May 29, 13
32. Say that 5 times fast. PEARSON Product moment correlation coefficient Wednesday, May 29, 13
33. 0 1.5 3 4.5 6 0 225 450 675 900 CPU vs Response Time Wednesday, May 29, 13
34. 0 1.5 3 4.5 6 0 225 450 675 900 CPU vs Response Time Wednesday, May 29, 13
35. Wednesday, May 29, 13
36. 1 data = [ 2 [100, 0.7], 3 [125, 0.5], 4 [150, 1], 5 [300, 2.1], 6 [500, 3.4], 7 [900, 6] 8 ] 9 10 x, y = data.transpose 11 n = data.size 12 x_mean = x.reduce(:+) / n 13 y_mean = y.reduce(:+) / n 14 x_stddev = Math.sqrt(x.inject {|sum, i| sum + (i - x_mean)**2 } / (n - 1).to_f) 15 y_stddev = Math.sqrt(y.inject {|sum, i| sum + (i - y_mean)**2 } / (n - 1).to_f) 16 z_x = x.collect {|i| (i - x_mean) / x_stddev } 17 z_y = y.collect {|i| (i - y_mean) / y_stddev } 18 pearsons = z_x.zip(z_y).collect {|x| x[0] * x[1] }.reduce(:+) / n 19 # => 0.9265763490538744 Wednesday, May 29, 13
37. PEarson Wednesday, May 29, 13
38. PEarson • Close to absolute 1 = probably correlated samples Wednesday, May 29, 13
39. PEarson • Close to absolute 1 = probably correlated samples • Could be applied to moving averages? Wednesday, May 29, 13
40. PEarson • Close to absolute 1 = probably correlated samples • Could be applied to moving averages? • Could we pull it into a graphite function? (Hackathon anyone?) Wednesday, May 29, 13
41. LIMITS OF Mathematical correlation Wednesday, May 29, 13
42. LIMITS OF Mathematical correlation • Requires known inputs and assumptions Wednesday, May 29, 13
43. LIMITS OF Mathematical correlation • Requires known inputs and assumptions • Suggestion of correlation, not proof Wednesday, May 29, 13
44. LIMITS OF Mathematical correlation • Requires known inputs and assumptions • Suggestion of correlation, not proof • Needs a large amount of knowledge of the data set to make decisions Wednesday, May 29, 13
45. I can see it! TIME BASED VISUAL CORRELATION Wednesday, May 29, 13
46. A Graphite Story Wednesday, May 29, 13
47. LIMITS OF VISUAL CORRELATION Wednesday, May 29, 13
48. LIMITS OF VISUAL CORRELATION • Takes a good eye Wednesday, May 29, 13
49. LIMITS OF VISUAL CORRELATION • Takes a good eye • Hard to see the signal through the noise Wednesday, May 29, 13
50. LIMITS OF VISUAL CORRELATION • Takes a good eye • Hard to see the signal through the noise • Doesn’t really account for domino events Wednesday, May 29, 13
51. LIMITS OF VISUAL CORRELATION • Takes a good eye • Hard to see the signal through the noise • Doesn’t really account for domino events • Good for trends but not as much for events Wednesday, May 29, 13
52. There’s a disturbance in the force. EMOTIONAL CORRELATION Wednesday, May 29, 13
53. RASHoMONING Wednesday, May 29, 13
54. Each person uses their unique knowledge of the situation to point out unique data points. RASHoMONING Wednesday, May 29, 13
55. LIMITS OF EMOTIONAL correlation Wednesday, May 29, 13
56. LIMITS OF EMOTIONAL correlation • Provides a trail not an answer Wednesday, May 29, 13
57. LIMITS OF EMOTIONAL correlation • Provides a trail not an answer • Depends on having a team of people Wednesday, May 29, 13
58. LIMITS OF EMOTIONAL correlation • Provides a trail not an answer • Depends on having a team of people • Many ideas, needs a “judge” Wednesday, May 29, 13
59. LIMITS OF EMOTIONAL correlation • Provides a trail not an answer • Depends on having a team of people • Many ideas, needs a “judge” • HUMANS (Hence Rashomoning) Wednesday, May 29, 13
60. I’m sold, show me how! Wednesday, May 29, 13
61. But I have some ideas. I don’t know exactly Wednesday, May 29, 13
62. TRYING TO MAKE THE DATA MORE VISIBLE Wednesday, May 29, 13
63. A conflagration of data HOTPOT Wednesday, May 29, 13
64. Wednesday, May 29, 13
65. Hotpot = Chef, Sensu, Graphite, (Logstash) Simply align disparate sources of data TO VISUALLY CORRELATE Wednesday, May 29, 13
66. create relationships for alerts and notifications Wednesday, May 29, 13
67. Wednesday, May 29, 13
68. Math to filter out noise. USe PEARSONS to pull out potentially related data Wednesday, May 29, 13
69. Have the ability to easily divide datasets by “cohorts” cohort analysis for processes/nodes Wednesday, May 29, 13
70. “Node notes”. Document everything. Treat personal and institutional knowledge as data Wednesday, May 29, 13
71. By making more data available to everyone. Make emotional correlation less EMO Wednesday, May 29, 13
72. So if that’s just level 2, what’s level 3? Wednesday, May 29, 13
73. Wednesday, May 29, 13
74. TAKE the correlations and let the machine turn them into decisions Wednesday, May 29, 13
75. Lets figure out level 2 first. NOT YET. Wednesday, May 29, 13
76. github.com/quirkey github.com/paperlesspost twitter.com/aq twitter.com/paperlessdev quirkey.com paperlesspost.com THANKS! Wednesday, May 29, 13