Evolutionary Architectures

This talk was given at the TLP Innova 2018.

One of the most difficult things in the modern development is tackling the complexity. Based in the ideas proposed by Neal Ford, Rebecca Parsons, and Patrick Kua in their book, the slides contains some techniques to help to reduce the complexity during the software development process, based in my own experience.

Transcript

1. Evolutionary Architectures Yeray Darias @ydarias
2. Talia Gershon Ismael Faro Paco Martín Fran Cabrera Jorge Carballo Jorge Barroso Abdón Rodríguez Iván Durán Jesús Pérez Toni Martínez Juan Gómez Axel Hernández Diego Moreda Juan Cruz-Benito Antonio López Yeray Darias IBM Research ETX David García Salva de la Puente Luciano Bello Krissy Brimijoin Borja Godoy Hyung Seo Greg Boland Olivia Ruan Joseph Schwarz Ismael Rozas
3. Act 1: Final Boss
6. Time Complexity
7. Time Complexity
10. Monolith Microservices Serverless
11. Monolith Microservices Serverless
15. Act 2: Starting the adventure
17. 1. One Page Design
23. Anti-pattern
24. 1. One Page Design Pros • It is easy • Keep all the team in the same page Cons • Harder to evolve and maintain than code References • Original talk by Stone Librande (link) • Talk by Jorge Barroso (link)
25. 2. Light. Architectural Decisions Record
27. Anti-pattern
28. 2. Light. Architectural Decisions Record Pros • It is easy • Help the team to remember why they took some decisions • Updated with the source code Cons • People tend to forget to update the records References • Entry at the ToughtWorks Technology Radar (link)
29. 3. Postmortems
32. Anti-pattern
33. 3. Postmortems Pros • Easy enough to write • Help the team to improve and avoid past mistakes Cons • Some people use them to blame References • GitLab Postmortem after its January 31 outage (link) • Runnics Postmortem slides (link)
34. 4. Continuos delivery
35. Anti-pattern
36. 4. Continuos delivery Pros • Lower lead time to production • Usually it provides a very professional attitude of the team Cons • Hard to implement • Serious bugs can go into production environments References • Martin Fowler talk about Continuos Delivery (link)
37. 5. Infrastructure as code
38. Anti-pattern
39. 5. Infrastructure as code Pros • New environments are deployed in minutes • The infrastructure documentation is up to date, because it is code! Cons • Very hard to implement • Usually it creates some information silos around the team
41. Act 3: Learning new ways to fight
43. Evolutionary Architecture
44. Architecture
47. Anti-pattern
49. Evolutionary
51. Evolutionary Architecture Small changes Software dimensions Fitness functions
52. Software dimensions (-ilities)
53. accessibility accountability accuracy adaptability administrability affordability agility auditability autonomy availability compatibility composability configurability correctness credibility customizability debugability degradability determinability demonstrability dependability deployability discoverability distributability durability effectiveness efficiency usability extensibility failure transparency fault tolerance fidelity flexibility inspectability installability integrity interoperability learnability maintainability manageability mobility modifiability modularity operability orthogonality portability precision predictability process capabilities producibility provability recoverability relevance reliability repeatability reproducibility resilience responsiveness reusability robustness safety scalability seamlessness self- sustainability serviceability securability simplicity stability standards compliance survivability sustainability tailorability testability timeliness traceability
54. accessibility accountability accuracy adaptability administrability affordability agility auditability autonomy availability compatibility composability configurability correctness credibility customizability debugability degradability determinability demonstrability dependability deployability discoverability distributability durability effectiveness efficiency usability extensibility failure transparency fault tolerance fidelity flexibility inspectability installability integrity interoperability learnability maintainability manageability mobility modifiability modularity operability orthogonality portability precision predictability process capabilities producibility provability recoverability relevance reliability repeatability reproducibility resilience responsiveness reusability robustness safety scalability seamlessness sustainability serviceability securability simplicity stability standards compliance survivability sustainability tailorability testability timeliness traceability
55. accessibility accountability accuracy adaptability administrability affordability agility auditability autonomy availability compatibility composability configurability correctness credibility customizability debugability degradability determinability demonstrability dependability deployability discoverability distributability durability effectiveness efficiency usability extensibility failure transparency fault tolerance fidelity flexibility inspectability installability integrity interoperability learnability maintainability manageability mobility modifiability modularity operability orthogonality portability precision predictability process capabilities producibility provability recoverability relevance reliability repeatability reproducibility resilience responsiveness reusability robustness safety scalability seamlessness sustainability serviceability securability simplicity stability standards compliance survivability sustainability tailorability testability timeliness traceability
57. Availability Resiliency Responsiveness Traceability Scalability Affordability
58. Availability Resiliency Responsiveness Traceability Scalability Affordability
59. Availability Resiliency Responsiveness Traceability Scalability Affordability
60. Availability Resiliency Responsiveness Traceability Scalability Affordability
61. Fitness Functions
63. f(x) = 56
64. f(x) = 79
65. f(x) = 43
66. Atomic Holistic Triggered Continuos Unit tests Legal audit Load test
67. Act 4: Ready to fight
69. 6. Static code analysis
70. Anti-pattern
71. 6. Static code analysis Pros • There are a lot of tools to implement it • Helps to have a better and safer code • Advanced tools can verify aspects like security or concurrency problems Cons • It is not magic, requires an effort from developers References • Measure and improve code quality (link) • Detecting and analyzing redundant code (link)
72. 7. Architectural rules enforcement
73. Anti-pattern
74. 7. Architectural rules enforcement Pros • Helps to organize the code better • It helps to keep the system more “modular” Cons • It requires a great effort to have some common understanding References • Architecture blog post by @herbertograca (link)
75. 8. Monitoring
76. Anti-pattern
77. 8. Monitoring Pros • Helps to know how each change affects the overall system Cons • Quite difficult in distributed architectures References • Mastering Chaos - a Netflix presentation (link) • Monitoring Microservices by Adrian Crockford (link)
78. 9. Social code analysis
80. Anti-pattern
81. 9. Social code analysis Pros • It helps to localize hot spots of code Cons • Very dependent on the analysis tool • Hard to implement from scratch References • Analyzing software with forensic psychology by Adam Tornhill (link)
83. One change at a time
84. Decide the most important dimensions for you
85. Measure how your changes affect the system
86. Automatic!!!
87. Q & A