4 Questions to Ask Your Dev Team

Good collaboration between Platform Teams and Application Teams is critical for delivering any production service. Practices such as site reliability engineering and continuous delivery must be understood by everyone to be effective, but creating that shared understanding can be hard in an existing organization. In this talk, Hannah and Jérôme will provide a simple guide to getting started.

Our experience has shown us that Platform Teams and Application Teams don’t talk enough, and we don’t talk about the right things. This session will give you four simple questions to start that dialogue—you might be surprised what you learn if you ask the right questions!

This talk was first given at SpringOnePlatform Oct 2019.

Transcript

1. © Copyright 2019 Pivotal Software, Inc. All rights Reserved. Hannah Foxwell Jérôme Wiedemann 8th October 4 Questions to Ask Your Dev Team
2. Why would I want to talk to developers? @HannahFoxwell @romrider42
3. Everything takes too long! Devs don’t care about production! @HannahFoxwell @romrider42
4. They don’t know what they need! I can do this cloud thing myself! @HannahFoxwell @romrider42
5. Who are we? Hannah Foxwell Jérôme Wiedemann Associate Director for Pivotal Labs Platform Services EMEA Associate Director for Pivotal Labs Platform Services EMEA We’re on a mission to build wildly successful Platform Teams. Technology is the easy part. People are the hard part! @HannahFoxwell @romrider42
6. Question 1: How are we going to work together? @HannahFoxwell @romrider42
7. Good fences make good neighbours. @HannahFoxwell @romrider42
8. IaaS Platform Application Platform Team Mission: “Build an amazing product and valuable service for my users!” Dev Ops Application Team Mission: “Build an amazing product and valuable service for my users!” Infrastructure Team Mission: “Build an amazing product and valuable service for my users!” @HannahFoxwell @romrider42
9. Some other patterns... Platform Eng AppDev WARNING: Only if you absolutely have to. Avoid silos with good communication and collaboration. Dev Ops AppOps Platform Ops Platform Application Dev Ops Tools Platform Application Dev Ops SRE Too much centralisation can become a constraint, but providing self serve tools can help developers focus on what’s important! SRE can be a separate team or it can be a set of practices adopted by everyone. @HannahFoxwell @romrider42
10. I Need More @HannahFoxwell @romrider42
11. The Platform is a Product Story 1 Story 3 Story 2 Story 1 Story 3 Story 2 Application Users Application Developers Platform Users Product Manager Platform Engineers Product Manager Think of application developers as your customers! @HannahFoxwell @romrider42
12. Question 2: What is your path to production? How can I help make it easier? @HannahFoxwell @romrider42
13. "Production is the happiest place on earth" Josh Long - @starbuxman @HannahFoxwell @romrider42
14. Production is where we deliver value to our users @HannahFoxwell @romrider42
15. So why are we so scared of production? @HannahFoxwell @romrider42
16. DORA's Aspect of Delivery Performance Source: https://cloud.google.com/devops/ @HannahFoxwell @romrider42
17. Automation helps you build reproducible workflows @HannahFoxwell @romrider42
18. Test Driven Development helps you build confidence @HannahFoxwell @romrider42
19. Safely release code by including feature toggles @HannahFoxwell @romrider42
20. Question 3: What level of reliability do our users need? @HannahFoxwell @romrider42
21. Reliability is Fundamental @HannahFoxwell @romrider42
22. Reliability is Everyone’s Goal @HannahFoxwell @romrider42
23. Reliability is a Team Sport @HannahFoxwell @romrider42
24. Service Level Indicators (SLIs), Service Level Objectives (SLOs) and Error Budgets @HannahFoxwell @romrider42
25. Service Level Indicator (SLI) is a metric Service Level Objective (SLO) is a threshold @HannahFoxwell @romrider42
26. SLO Error Budget (per 28 Days) 99% 403 mins 99.5% 202 mins 99.9% 40.3 mins 99.95% 20.2 mins 99.99% 4.03 mins 99.999% 0.04 mins Your Error Budget is the inverse of your SLO @HannahFoxwell @romrider42
27. Who gets to use the Error Budget? @HannahFoxwell @romrider42
28. Sharing Responsibility for Users Happiness IaaS Platform Application SLO = 99.99% Error Budget = 4.03 minutes SLO = 99.95% Error Budget = 20.2 minutes SLO = 99.9% Error Budget = 40.3 minutes @HannahFoxwell @romrider42
29. Agree an Error Budget Policy before you need one @HannahFoxwell @romrider42
30. Example Error Budget Policy Service is performing at or above SLO Service has exceeded Error Budget in the preceding 4 week window A single incident consumes more than 20% of Error Budget over 4 weeks Continue to release changes as normal within agreed release policy Halt all changes and releases other than P1 and P2 issues or security fixes Post-incident review. Must include at least 1 action to address the cause @HannahFoxwell @romrider42
31. Question 4: What are we going to do when things go wrong? @HannahFoxwell @romrider42
32. 100% availability is NOT the right expectation @HannahFoxwell @romrider42
33. Only actionable alerts should trigger incident response @HannahFoxwell @romrider42
34. Replicate user interactions @HannahFoxwell @romrider42
35. Build confidence with fire drills @HannahFoxwell @romrider42
36. Blameless incident retrospective @HannahFoxwell @romrider42
37. So in summary... @HannahFoxwell @romrider42
38. Talk to each other @HannahFoxwell @romrider42
39. Ask these 4 simple questions: 1. How are we going to work together? 2. What is your path to production? 3. What level of reliability do our users need? 4. What are we going to do when things go wrong? Let’s work together! @HannahFoxwell @romrider42
40. Are we getting better at this? Ask these 4 simple questions: 1. How are we going to work together? 2. What is your path to production? 3. What level of reliability do our users need? 4. What are we going to do when things go wrong? @HannahFoxwell @romrider42
41. Transforming how the world builds software © Copyright 2019 Pivotal Software, Inc. All rights Reserved.