testing in devops: ukstar conference

22
Test value first before correctness 27-28 February 2017 LONDON Laurent Py @py_laurent [email protected] et https://hiptest.net

Upload: laurent-py

Post on 13-Apr-2017

79 views

Category:

Software


3 download

TRANSCRIPT

Page 1: Testing in DevOps: UKStar conference

Test value first before correctness

27-28 February 2017LONDON

Laurent Py

@[email protected]://hiptest.net

Page 2: Testing in DevOps: UKStar conference

PLAN

2

The match: quality VS speed

Step 1: Testing the idea

Step 2: Testing first increment in production

Step 3: General availability of the feature

Benefits of a progressive approach to quality

Page 3: Testing in DevOps: UKStar conference

27-28 February 2017LONDON

QUALITY VS SPEED

Page 4: Testing in DevOps: UKStar conference

WE HAVE SPEED

4

Between 5 to 10 deployments/day

Our DevOps

teamDevs/testers

Ops

Product manager

Users

Marketing

Support

4

Page 5: Testing in DevOps: UKStar conference

WHAT ABOUT THE USER?

5

Speed of development feedback and learning

Value VS Quality

5

Page 6: Testing in DevOps: UKStar conference

27-28 February 2017LONDON

STEP 1: TESTING THE IDEA

Page 7: Testing in DevOps: UKStar conference

8

• Use personae • Capture the expected behavior with BDD • Define business assumptions

Benefits

Support

Devs/testeurs

Marketing

Ops

Product manager

WHAT VALUE DO WE WANT TO DELIVER?

8

Page 8: Testing in DevOps: UKStar conference

9

• Create a shared understanding with examples• Based on a common business language• Definition of stop• When tests are automated → living

documentation

BDD AT A GLANCE

9

Page 9: Testing in DevOps: UKStar conference

1111

Page 10: Testing in DevOps: UKStar conference

27-28 February 2017LONDON

STEP 2: TESTING FIRST INCREMENT IN PRODUCTION

Page 11: Testing in DevOps: UKStar conference

13

1. After step 1, we have our definition of success

2. We develop a first increment, do ET

3. We deploy it in production

FEATURE IS DEVELOPED

13

Page 12: Testing in DevOps: UKStar conference

14

• Based on usage- NB users, clicks- NB requests…

• Impact on- Sign-up- Revenue- Retention…

LET’S MEASURE THE IMPACT

14

Page 13: Testing in DevOps: UKStar conference

15

EXAMPLES OF METRICS

15

Notification feature• Usage

- NB clicks NB of notifications- NB of users

• Business: none

Test refactoring• Usage

- NB of refactoring actions- NB of impacts

• Business: Retention

Page 14: Testing in DevOps: UKStar conference

16

LIVE CHAT TO GET REAL TIME FEEDBACK

16

Page 15: Testing in DevOps: UKStar conference

1717

Page 16: Testing in DevOps: UKStar conference

27-28 February 2017LONDON

STEP 3: FEATURE FULLY AVAILABLE

Page 17: Testing in DevOps: UKStar conference

19

1. The impact has been validated

2. Investment on performance and non regression (checks are fully automated)

3. Feature made available to all users with a switch

FEATURE BECOMES FULLY AVAILABLE

19

Page 18: Testing in DevOps: UKStar conference

20

Appsignal, Pingdom, Scalingo et Logmatic

MONITORING PERFORMANCE AND BUSINESS

20

Page 19: Testing in DevOps: UKStar conference

2121

Page 20: Testing in DevOps: UKStar conference

27-28 February 2017LONDON

CONCLUSION

Page 21: Testing in DevOps: UKStar conference

23

•Test the value first• Invest progressively on quality•Create intimacy with the users•Create a suitable environment for experimentations

KEY TAKEAWAYS

23

Page 22: Testing in DevOps: UKStar conference

Thank you!

27-28 February 2017LONDON

https://hiptest.net