SLO Driven Development
•
0 likes•80 views
This document discusses service level objectives (SLOs) and how they can be used to improve system reliability. Some key points: - SLOs define reliability targets for services based on user expectations rather than strict SLAs, allowing for more flexibility. - Service level indicators (SLIs) are used to measure how well an objective is being met by tracking key metrics. - SLOs can be used to establish an "error budget" that accounts for expected failures and drives improvement by spending the budget on tests, monitoring, etc. - Setting appropriate SLOs and SLIs involves tradeoffs between development speed, maintenance costs, and user experience. Data is crucial to defining and optimizing
Report
Share
SLO Driven Development
- 3. SLO DRIVEN DEVELOPMENT Photo by Oscar Keys on Unsplash
- 4. Name: Alon Nativ Company: Tomorrow.io Hobbies: Rant about Python @anativ /anativ
- 5. Accurate ML & Data Save Lives
- 7. LIGHTNING ALERT Image by Keli Black from Pixabay
- 8. Notify on Every Lightning In Less Than 500 ms Image by David Schwarzenberg from Pixabay
- 12. Notify on Every Lightning In Less Than 500 ms Image by David Schwarzenberg from Pixabay
- 13. Notify on EVERY Lightning In Less Than 500 ms Image by David Schwarzenberg from Pixabay
- 14. The most IMPORTANT feature of any system is its RELIABILITY
- 15. 100% is the wrong RELIABILITY target for basically EVERYTHING. Benjamin Trynor Sloss VP of 24x7 Engineering, Google
- 16. every number that you pick has DIRECT IMPACT on your cost velocity and architecture 99.5 99.4 99.6 99.7 99.8 99.9 1 Image by Arek Socha from Pixabay
- 17. SLA Photo by Todd Quackenbush on Unsplash
- 18. SLA - Service Level Agreement Binding Agreement Pay Sales & Customers Image by Okan Caliskan from Pixabay
- 19. DOWN OVER a DAY Image by Markus Kammermann from Pixabay
- 20. SLA Refund Time / Month 99% > X >= 95% 25% 1d 12h 31m
- 21. If you are proud of your SLA you are probably doing something WRONG
- 22. SLA SLO Photo by Todd Quackenbush on Unsplash
- 23. SLO - Service Level Objectives User Happiness Your Expectations Product & SRE*
- 25. SLA SLO SLI SLI SLI Photo by Todd Quackenbush on Unsplash
- 26. SLI - Service Level Indicator Key Metrics Monitors Developers & SRE*
- 27. SLI = good events valid events X 100
- 29. GOOD SLI Up to 4 No Internal* Metrics 2-4
- 31. HIGH CORRELATION
- 33. This slide can’t be reached ERROR_NO_SLIDE_FOUND
- 34. 1 - SLO = ERROR BUDGET
- 36. SPENDING ERROR BUDGET SLI Error Budget
- 37. SPENDING ERROR BUDGET SLI Error Budget
- 39. MetaGoat
- 40. Team A: TESTS Team B: CI/CD
- 41. W. Edwards Deming. Data Scientist Without DATA you are another person with an OPINION.
- 42. MTTR / MTTF
- 43. MTTR Mean Time To Recovery
- 44. MTTF Mean Time to Failure
- 45. Team B: MTTR (rollback) Team A: MTTF (tests)
- 46. Team A: MTTF (tests)
- 47. Team B: MTTR (rollback)
- 48. Team B: MTTR (rollback) Team A: MTTF (tests)
- 49. If you can’t MEASURE it, you can’t IMPROVE it. Lord Kelvin Mathematician & engineer
- 50. TRADEOFFS
- 51. SPARE BUDGET Features Risky Experiments Spot / preemptible Scale Down A/B Testing Image by PublicDomainPictures from Pixabay
- 52. OUT OF BUDGET Deployment freeze Post Mortem CI/CD Monitoring Relax SLO Deprecate Services Photo by Allef Vinicius on Unsplash
- 53. HIGH SLO Less Budget Development Time Sleeping Hours Maintenance x2-x10 Photo by Gina Neri on Unsplash
- 54. FIRST STEPS Image by Pexels from Pixabay
- 55. USER CENTRIC Photo by Mark Pan4ratte on Unsplash
- 56. In GOD we trust all others bring DATA W. Edwards Deming. Data Scientist Photo by Jackson David on Unsplash
- 57. USE YOUR BUDGET Image by Olya Adamovich from Pixabay
- 58. Users Data Budget Photo by montatip lilitsanong on Unsplash
- 59. If you can’t manage your RELIABILITY, your reliability MANAGES you Photo by Jonathan Klok on Unsplash