How to use SLOs to understand user experience and find out how to make them happy @ TECHPULSE 2023

- Speaker: Oreo Liou
- Event: http://techpulse.line.me/

過去在服務發生異常時，商業團隊只能確認服務狀態，但因不了解用戶狀態，團隊將很難做出正確的決策以改善服務。

另一方面，研發和商業團隊對於服務的定義可能有所不同，也會導致團隊之間無法清楚溝通，無法決定是否投入成本。

為了克服這些挑戰，今天將與大家分享如何從觀察指標到理解用戶體驗的方法，分享如何衡量和定義服務水平，讓團隊能夠做出更好的決策並改善服務品質。

Transcript

1. 1
   How to use SLOs to understand user experience and
   fi
   nd out how to make them happy?
   

   View Slide

2. Background
   2
   What happened ?
   2
   Normal Crash
   Biz
   Dev Pm
   Service ?
   

   View Slide

3. Normal
   3
   How To Measure Quality of Service ?
   3
   Normal
   ｘｃ
   Biz
   Dev Pm
   Background
   

   View Slide

4. Observability Platform
   4
   Observability platform collects Metrics, Log, Tracing data.
   

   View Slide

5. Service Level Objectives
   5
   Since Google
   fi
   rst introduced Service Level Indicators
   (SLIs) and Service Level Objectives (SLOs) as part of
   its Site Reliability Engineering (SRE) practice
   Observational tools such as Datadog, Honeycomb
   have long since incorporated SLO monitoring into their
   toolchains.
   OpenSLO is A declarative, neutral speci
   fi
   cation
   language emerging based on the YAML format used by
   Kubernetes and allowing SLIs and SLOs to be de
   fi
   ned
   as code standard.
   

   View Slide

6. 6
   SLA
   SLO
   SLI
   Promise
   Objectives
   Measurement
   The agreement between
   client & user
   The objective a service
   Provider must achieve to
   Meet the agreement
   The real number of per
   Performance. How did the
   Service provider do this job
   Service Level Agreements
   Service Level Objectives
   Service Level Indicators
   

   View Slide

7. SLA / SLO / SLI
   7
   User Experience
   ？
   

   View Slide

8. SLA / SLO / SLI
   8
   

   View Slide

9. 4 ⽉
   5 ⽉
   6 ⽉
   7 ⽉
   0 25 50 75 100
   9
   Now we have SLOs, but biz still not happy?
   SLO 1

   SLO 2

   SLO 3
   ….
   SLO 100
   

   View Slide

10. CUJ
    10
    Biz
    Live
    …
    …
    …
    

    View Slide

11. 4 steps to identify and implement effective SLOs
    Understand internal
    business goals, and
    external
    stakeholders
    Identify and
    prioritize critical
    services that affect
    SLAs
    Identify key metrics
    to use as service-
    level indicators
    (SLIs)
    Identify key SLOs
    SLA SLI SLO
    CUJ
    11
    

    View Slide

12. Hold a SLOs workshop
    12
    

    View Slide

13. Define SLOs document
    13
    

    View Slide

14. SLOs Spec Yaml
    14
    

    View Slide

15. CRDs
    SLOs generator
    Prometheus
    Rule
    Alert Rule
    SLI Rule
    CRDs
    CUJ & SLOs
    Spec
    15
    K8s
    Grafana
    

    View Slide

16. SLOs Dashboard
    16
    

    View Slide

17. SLOs Timeline
    17
    

    View Slide

18. Q1 Q2 Q3 Q4
    Project A (1/1) (1/1) (1/1) (1/1)
    Project B (1/1) (1/1) (1/1) (1/1)
    Project C (1/1) (1/1) (1/1) (1/1)
    Project D (5/5) (7/10) (9/10) (10/10)
    SLOs Report
    18
    

    View Slide

19. Continuously improve your SLOs
    19
    SLO Dashboard Weekly meeting Jira ticket system
    

    View Slide

20. Takeaway
    20
    SLA
    SLO
    SLI
    Promise
    Objectives
    Meeasurement
    Service Level Agreements
    Service Level Objectives
    Service Level Indicators
    20
    Build your observation system
    Discuss goals with your team
    Keep your promise
    

    View Slide

21. Thank you
    21
    

    View Slide