Videos / DevOps

Episode 11 of 14 · DevOps, end to end

Monitoring and observability: knowing what is happening, and why

Chapters

1. The wall of green0:00
2. The question you did not ask0:15
3. The working test0:32
4. Three signals, three jobs0:47
5. Cardinality picks the questions1:06
6. Emit it once, one shape1:26
7. The odometer and the speedometer1:44
8. The gauge, and what it misses2:01
9. Counting in buckets2:20
10. The average is a fiction2:40
11. Percentiles instead2:59
12. What to record, and how much3:17
13. The line written for a person3:37
14. Fields, not sentences3:53
15. One wide event per unit4:12
16. One id, threaded through4:31
17. Keep the interesting ones4:49
18. Ingest, index, keep5:07
19. One request, itemised5:23
20. A tree of spans5:41
21. The id has to travel5:57
22. The waterfall6:15
23. Shapes an average hides6:33
24. Decide after, not before6:50
25. Up is not a number7:10
26. Measure where the user is7:28
27. A target, and a window7:45
28. What is left over8:03
29. Burn rate8:20
30. What the budget buys you8:38
31. The page at three am8:56
32. Symptoms, not causes9:14
33. Three tests before waking9:31
34. Fast burn, and slow burn9:49
35. The runbook attached10:07
36. Review every page you got10:24
37. The whole thing10:42

Monitoring and observability: knowing what is happening, and why. Every dashboard on the wall is green. The graphs are flat, the checks pass, and the support queue is filling with people who cannot check out. Nothing is broken, according to anyth

Mark as watchedSaved in this browser only.