The Future of Developer Careers: When Dev and Ops Meet

The last decade and change has seemingly been all about developer productivity: shipping code faster, delivering continuously, and breaking apart monoliths in the name of greater flexibility. But the next frontier yet looms: production. Leading software engineering organizations are increasingly asking developers to own their code in production—but what does that mean, in practice? We'll discuss tooling (ready to embrace developers, or built for a different audience?), culture (how does embracing production help us skill up as devs?), and more in this talk.

Transcript

1. The Future of Developer Careers When Dev and Ops Meet @cyen @honeycombio
2. I’m Christine, and I’m a dev. Hi there. I’m Christine
4. DEV PROD DEV OPS
5. DEV PROD DEV OPS
6. DEV OPS DEV PROD
7. DEV OPS DEV PROD
8. DEV PROD DEV OPS observability
9. Leveling up at Production: the naive way
10. "Leaps of intuition akin to magic" ! " Sure, you just ັ፡㾿 ᤒ҅แ"ว$ ݺ۸ਸ ੑ۱ … could you repeat that? DEV OPS #
11. DEV
12. The Software ▸ Design documents ▸ Architecture review ▸ Test-driven development ▸ Integration tests ▸ Code review ▸ Continuous integration ▸ Continuous deployment ▸ $%&' ▸ Observe our code in production DEV Process TEST
13. --- FAIL: TestUnit (0.00s) talk_test.go:10: -- actual: 5 (type int) expected: 4 (type int) ACTUAL EXPECTED
14. --- FAIL: TestUnit (0.00s) talk_test.go:10: -- actual: 5 (type int) expected: 4 (type int) ACTUAL EXPECTED
15. (an aside)
16. Leveling up at Production: the right way
17. ( ‣Is my code running? ‣Is it behaving as expected in prod? DEV build_id commit_sha is_feature_flag_live
18. ( ‣Is my code running? ‣Is it behaving as expected in prod? DEV user_id functionality_tier time_zone api_endpoint fn_name
19. speak the language of devs CPU utilization AWS availability zone # of network connections Cassandra hostname build ID client OS API endpoint time to render $YOUR_BIZ-relevant ID DEV OPS kafka partition
20. share patterns where possible
21. The Software ▸ Design documents ▸ Architecture review ▸ Test-driven development ▸ Integration tests ▸ Code review ▸ Continuous integration ▸ Continuous deployment ▸ $%&' ▸ Observe our code in production DEV Process PROD
22. testing hypothesis full rollout (analysis, slicing, dicing)
23. ( ‣Is my code running? ‣Is it behaving as expected in prod? DEV ‣Is my code running? ‣Will it behave as expected in prod? what is normal?
24. "how bad is it, really?"
25. Practical steps forward
26. one step at a time change is incremental and iterative
27. ‣ start from the edge (breadth) ‣ ‣ ‣ one step at a time
28. ‣ start from the edge (breadth) ‣ focus on the pain (depth) ‣ ‣ one step at a time
29. one step at a time ‣ start from the edge (breadth) ‣ focus on the pain (depth) ‣ instrument from where you control ‣ APP USER DATABASE user_id endpoint params hostname active_queue request_dur_ms response_status_code ) query_sql caller_fn database_dur_ms num_rows_returned
30. ‣ start from the edge (breadth) ‣ focus on the pain (depth) ‣ instrument from where you control ‣ capture what matters to you (context) one step at a time
31. ‣ start from the edge (breadth) ‣ focus on the pain (depth) ‣ instrument from where you control ‣ capture what matters to you (context) one step at a time You (Individual) You (Team) You (Business)
32. ‣ start from the edge (breadth) ‣ focus on the pain (depth) ‣ instrument from where you control ‣ capture what matters to you (context) Your Infra - hostname - machine type Your Deploy - version / build - feature flags Your Business - customer - shopping cart Your Execution - payload characteristics - timers one step at a time
33. technical human ‣ distributed tracing ‣ high-cardinality data ‣ auto- instrumentation
34. technical human ‣ distributed tracing ‣ high-cardinality data ‣ auto- instrumentation ‣ iterate to improve ‣ explore! be curious! ‣ embed into process
35. So just remember…
38. Questions? play.honeycomb.io @cyen