What Can UX Designers Teach Us About Being Better Devops Practitioners?

This is a set of slides from "What Can UX Designers Teach Us About Being Better Devops Practitioners?" from Agile + Devops West in June 2020.

Product for Internal teams: https://medium.com/@skamille/product-for-internal-platforms-9205c3a08142 Fundamentals of Design: How to Think Like A Designer https://www.skillshare.com/classes/Fundamentals-of-Design-How-to-Think-Like-a-Designer/1986357063 Don’t Make Me Think: http://sensible.com/dmmt.html
Rocket Surgery Made Easy: http://sensible.com/rsme.html Bootstrap: https://getbootstrap.com/
Ant.Design: https://ant.design/
Design Sprints: https://www.gv.com/sprint/

Transcript

5. What Can UX Designers Teach Us About Being Better Devops Practitioners? Jonathan Maltz - June 11th, 2020
6. 6 Agenda Culture Process Tools
7. 7 Agenda Culture Tools
8. Culture
9. Build a culture of talking to users
12. Infrastructure teams build products.
13. 13 What Are Your Products? Jenkins Kubernetes Pagerduty Splunk
14. 14 Jenkins Kubernetes Pagerduty Splunk A product to help users reliably run CI jobs What Are Your Products?
15. 15 Jenkins Kubernetes Pagerduty Splunk A product to help users reliably run CI jobs A product to help users deploy + scale applications What Are Your Products?
16. 16 Jenkins Kubernetes Pagerduty Splunk A product to help users reliably run CI jobs A product to help users deploy + scale applications A product to help users respond to incidents What Are Your Products?
17. 17 Jenkins Kubernetes Pagerduty Splunk A product to help users reliably run CI jobs A product to help users deploy + scale applications A product to help users respond to incidents A product to help users debug failures in their system What Are Your Products?
18. 18 Who are your users?
19. 19 Who are your users? Name: Just-Get-It-Done Jonathan Role: Software engineer Background: - Previously worked as a mobile engineer - Now building services for risk operators at Stripe - Occasionally goes to devops conferences Goals: - Get his services deployed easily - Help his team effectively use infra tools
20. 20 These products have interfaces apiVersion: apps/v1 kind: Deployment metadata: name: nginx-deployment spec: selector: matchLabels: app: nginx replicas: 2 template: metadata: labels: app: nginx spec: containers: - name: nginx image: nginx:1.14.2 ports: - containerPort: 80
21. 21 These products have interfaces apiVersion: apps/v1 kind: Deployment metadata: name: nginx-deployment spec: selector: matchLabels: app: nginx replicas: 2 template: metadata: labels: app: nginx spec: containers: - name: nginx image: nginx:1.14.2 ports: - containerPort: 80 kubectl apply service.yaml
22. 22 These products have interfaces service_image: stripe/test-service instance_size: r2.xlarge min_instances: 1 max_instances: 100
23. 23 These products have interfaces service_image: stripe/test-service instance_size: r2.xlarge min_instances: 1 max_instances: 100
24. Your users are all around you.
25. Your users are all around you. Talk to them
27. 27 How to Talk to Users Find users Ask them questions Synthesize to make changes Repeat
28. 28 How to Talk to Users Find users Ask them questions Synthesize to make changes Repeat
29. Get diversity
30. Get diversity of role
31. Get diversity of experience
32. Get diversity of background
33. 33 How to Talk to Users Find users Ask them questions Synthesize to make changes Repeat
34. Ask open ended questions
35. “Do you like using Jenkins for testing services?”
36. “Do you like using Jenkins for testing services?”
37. “Can you talk to me about how you interacted with Jenkins on your last PR?”
38. “Can you talk to me about how you interacted with Jenkins on your last PR?” ✅
39. Users
40. They are not
41. They are not wrong
42. 42 How to Talk to Users Find users Ask them questions Synthesize to make changes Repeat
43. “I’m not really sure what that does. Whenever I have a CI failure I have to spend 3 minutes clicking around to find the stack trace ”
44. “I’m not sure which buttons I can safely click here”
45. “I often keep a few branches running because CI takes a long time.”
46. Themes 46
47. Themes Speed of results 1 47
48. Themes Speed of results Simple error reporting 1 2 48
49. Themes Speed of results Simple error reporting Ability to dig in as required 1 2 49 3
54. 54 How to Talk to Users Find users Ask them questions Synthesize to make changes Repeat
57. Interview to understand needs
58. Interview to understand needs Decide on approaches
59. Interview to understand needs Decide on approaches Create lightweight mock-ups
60. Interview to understand needs Decide on approaches Create lightweight mock-ups
61. Interview to understand needs Decide on approaches Create lightweight mock-ups Decide on final approach
62. Interview to understand needs Decide on approaches Create lightweight mock-ups Decide on final approach Bug bash
63. Tools
65. 65 Major Principles Similarity Principle Proximity Principle Past Experience Principle Common Region
66. 66 Major Principles Similarity Principle Proximity Principle Past Experience Principle Common Region
67. 67 Similarity Principle
68. 68 Similarity Principle - In Practice Load Balancer Web Application Queue Workers Data Warehouse MySQL Queue Consumers
69. 69 Similarity Principle - In Practice Load Balancer Web Application Queue Workers Data Warehouse MySQL Queue Consumers
70. 70 Similarity Principle - In Practice Load Balancer Web Application Queue Consumers Data Warehouse MySQL Queue Workers
71. 71 Similarity Principle - In Practice Load Balancer Web Application Queue Consumers Data Warehouse MySQL Queue Workers
72. 72 Major Principles Similarity Principle Proximity Principle Past Experience Principle Common Region
73. 73 Proximity Principle
74. Queue worker throughput Overall error rate P99 response times 74 Proximity Principle - In Practice P99 response times Overall error rate Queue worker throughput Queue Size Underreplicated partitions Frontend error rate P50 response times Checkout fulfillment error rate P95 response times
75. 75 Proximity Principle - In Practice P99 response times Overall error rate Queue worker throughput Queue Size Underreplicated partitions Frontend error rate P95 response times P50 response times Checkout fulfillment error rate
76. 76 Past Experience Principle Similarity Principle Proximity Principle Past Experience Principle Common Region
77. 77 Past Experiences Principle
78. 78 Past Experience Principle - In Practice Are you sure you want to delete this graph? Yes No
79. 79 Past Experience Principle - In Practice Are you sure you want to delete this graph? Yes No
80. 80 Past Experience Principle - In Practice Are you sure you want to turn off this this cluster? Yes No
81. 81 Past Experience Principle Similarity Principle Proximity Principle Past Experience Principle Common Region
82. 82 Common Region
83. 83 Common Region
84. 84 Common Region
85. 85 Common Region - In Practice Your Deployments • Deployment #1: aaaaaa • Deployment #2: bbbbb • Deployment #3: cccccc All Deployments • Deployment #10: 123456 • Deployment #22: 987654 • Deployment #33: 024689
86. 86 Common Region - In Practice Your Deployments • Deployment #1: aaaaaa • Deployment #2: bbbbb • Deployment #3: cccccc All Deployments • Deployment #10: 123456 • Deployment #22: 987654 • Deployment #33: 024689
87. 87 Common Region - In Practice Your Deployments • Deployment #1: aaaaaa • Deployment #2: bbbbb • Deployment #3: cccccc All Deployments • Deployment #10: 123456 • Deployment #22: 987654 • Deployment #33: 024689
88. Design Systems
89. 89 Design Systems
90. Task: Build a tool to know if a service is healthy and create an incident if it’s not.
91. 91 What makes the service healthy? Responding reasonably quickly Users are seeing an acceptable level of errors Data is flowing to the data warehouse
92. Response times (p99) Response times (p95) Error Rates Checkout Errors Redshift Exporter Duration Redshift Delay (seconds) Create Incident
93. Response times (p99) Response times (p95) Error Rates Checkout Errors Redshift Exporter Duration Redshift Delay (seconds) Create Incident Open Open Open Open Open Open
94. Response times (p99) Response times (p95) Error Rates Checkout Errors Redshift Exporter Duration Redshift Delay (seconds) Create Incident Open Open Open Open Open Open
95. Response Latency Response times (p99) Error Rates Total Errors Redshift Connector Redshift Delay (seconds) Create Incident Healthy Healthy Unhealthy Open Open Open
96. Response Latency Response times (p99) Error Rates Total Errors Redshift Connector Redshift Delay (seconds) Create Incident Healthy Healthy Unhealthy Open Open Open
97. Response Latency Response times (p99) Error Rates Total Errors Redshift Connector Redshift Delay (seconds) Create Incident Open Open Open
98. Response Latency Response times (p99) Error Rates Total Errors Redshift Connector Redshift Delay (seconds) Create Incident Open Open Open
99. Response Latency Response times (p99) Error Rates Total Errors Redshift Connector Redshift Delay (seconds) Create Incident Open Open Open
100. Summary
101. Three Takeaways 101
102. Three Takeaways You’re building products. Your users are your colleagues 1 102
103. Three Takeaways You’re building products. Your users are your colleagues Involve your users in open-ended conversations 1 2 103
104. Three Takeaways You’re building products. Your customers are your colleagues Involve your customers in open-ended conversations You don’t need a design degree to make usable tools 1 2 104 3
106. Jonathan Maltz - @maltzj
107. Resources Product for Internal teams: https://medium.com/@skamille/product-for-internal-platforms-9205c3a08142 Fundamentals of Design: How to Think Like A Designer https://www.skillshare.com/classes/Fundamentals-of-Design-How-to-Think-Like-a-Designer/1986357063 Don’t Make Me Think: http://sensible.com/dmmt.html Rocket Surgery Made Easy: http://sensible.com/rsme.html Bootstrap: https://getbootstrap.com/ Ant.Design: https://ant.design/ Design Sprints: https://www.gv.com/sprint/ 107