DOO-004_楽天での DevOps 実践事例と Azure ベスト プラクティス
- 1. Enterprise Agile and DevOps Vol.01 May/25/2016 Yasunobu Kawaguchi Product Strategy Department, Rakuten Inc. http://tech.rakuten.co.jp/
- 2. 2 > whoami Yasunobu Kawaguchi Agile Coach
- 4. 4 Courtesy by Michael Sahota http://agilitrix.com/2015/03/enterprise-agile-agile-enterprise/
- 5. 5 Courtesy by Michael Sahota http://agilitrix.com/2015/03/enterprise-agile-agile-enterprise/
- 6. 6 A:チーム運営上の課題 A-1 プロジェクトチーム解散時の記憶喪失 A-2 チームの熟成不足 A-3 スキルとプロセス合意の不足 B:チーム外との調整の課題 B-1 チーム外との調整負担 B-2 承認プロセスが足を引っ張る B-3 予算管理や受発注契約と開発プロセスの不一致 C:組織全体の課題 C-1 評価制度のミスマッチ: C-2 指導者の不足 C-3 情報共有不足と不信感
- 8. 8 A-1 プロジェクトチーム解散時の記憶喪失 “When we ship a release at Microsoft, people often change jobs. For employees, this rotation is an opportunity both to develop a career and improve personal satisfaction in trying new challenges.” “Although this is a healthy pattern for the company and its employees overall, in the short term it can create a sort of amnesia.” Sam Guckenheimer, Neno Loje “Visual Studio Team Foundation Server 2012 : Adopting Software Practices” 「Microsoftでは製品リリースにこぎつけると通常は 人事異動が行われます。社員にとって、この人事異動は、 キャリアを開発し、新しい挑戦を行なって 個人の満足度を高める好機です。」 「これは企業にとっても社員全体にとっても 健全なやり方ですが、短期的には一種の 記憶喪失状態を生み出すことになります。」 「アジャイルソフトウェアエンジニアリング」 (日経BP社)
- 9. 9 A:チーム運営上の課題 A-1 プロジェクトチーム解散時の記憶喪失 A-2 チームの熟成不足 A-3 スキルとプロセス合意の不足 B:チーム外との調整の課題 B-1 チーム外との調整負担 B-2 承認プロセスが足を引っ張る B-3 予算管理や受発注契約と開発プロセスの不一致 C:組織全体の課題 C-1 評価制度のミスマッチ C-2 指導者の不足 C-3 情報共有不足と不信感
- 10. 10 A:チーム運営上の課題 A-1 プロジェクトチーム解散時の記憶喪失 A-2 チームの熟成不足 A-3 スキルとプロセス合意の不足 B:チーム外との調整の課題 B-1 チーム外との調整負担 B-2 承認プロセスが足を引っ張る B-3 予算管理や受発注契約と開発プロセスの不一致 C:組織全体の課題 C-1 評価制度のミスマッチ C-2 指導者の不足 C-3 情報共有不足と不信感 このあたり DevOpsが 効果ありそう
- 15. 15 What is ?
- 16. 16 Dev : Development Ops : Operations DevOps : Faster Feedback for Business
- 17. Old Way DevOps Way Automation Hand-offs and Approvals Small Team Reduce Waste (Time and Money)
- 22. 22 Microsoft Yahoo! Even in older companies…
- 23. 23 Rakuten
- 24. 24 2016 October
- 25. 25 Why?
- 26. 26 Agile Technical Excellence Small Team DevOps Cloud Enterprise Virtualization Automation Self Service Quality Organization Collaboration Customer Value and on
- 27. 27
- 28. 28 Hand-offs and Approvals Real Time Co-Work Enterprise Startups
- 29. 29 Hand-offs
- 32. 32
- 35. 35 Business Sales Creative Marketing Development DevOps Cross Functional Cover All Skillsets Trusted by others
- 36. 36
- 39. 39 Automation DevOps Enterprise Hand-offs and Approvals Self Service Fast and Frequent Value Stream
- 40. DevOps Real Life Examples at Rakuten Vol.01 May/25/2016 Dimitrov Kanio Ecosystem Service Department, Rakuten Inc. http://tech.rakuten.co.jp/
- 41. About Me Azure MVP, Architect & Advisor Tokyo Azure Meetup Host twitter: @azurekanio blog: https://azurekan.wordpress.com/
- 42. Black Ships 42
- 43. Rakuten Shugi 43 • Always Improve, Always Advance • Passionately Professional • Hypothesize -> Practice -> Validate -> Shikumika • Maximize Customer Satisfaction • Speed!! Speed!! Speed!!!
- 44. DevOps 44
- 45. Culture 45 Rakuten Shugi DevOps Always Improve, Always Advance Managing Technical Debt Passionately Professional Team Autonomy & Enterprise Alignment; Production First Mindset Hypothesize -> Practice -> Validate -> Shikumika Evidence gathered in Production; Backlog groomed with Learning Maximize Customer Satisfaction Flow of Customer Value Speed!! Speed!! Speed!!! Managing Infrastructure as a Flexible Resource
- 48. Process – Predictability and Discipline Dev Ops Team Tools Services Data Analysis Process
- 49. Process - Scrum
- 50. Team – Critical Success Factor Dev Ops Team Tools Services Data Analysis Process
- 51. Feature Teams Developers QA DevOps Product Manager Project Manager • Owns, Develops and Supports Features
- 52. Tools – DevOps Philosophy Applied Dev Ops Team Tools Services Data Analysis Process
- 53. Build Server - Roles • Central part of development process • Integrated with Messaging App (Hip Chat) forreal time reporting • Global Collaboration – Tokyo, Japan – San Mateo, USA – Cambridge, UK – Bangalore, India
- 54. Build Server - Roles • Centralized deployment • Configuration management • Resource provisioning in the cloud • Start / stop agents on demand
- 55. Build Server – Code Quality • Feature branches builds – Build during development • Reduce Code Review - Automation • Gated Builds – Merge only if successful build exist Build Unit Tests Code Coverage
- 56. Build Server – Code Quality Style Cop Sonar Cube Patch Config Backup Deployment Integration Tests Microsoft Azure
- 57. Build Server Set Up Build Server Agent AgentAgent Agent Tokyo San Mateo Cambridge Bangalore
- 58. Automation Tools • PowerShell • Azure Resource Manager • Azure Automation • Azure Backup
- 59. Cloud Services – optimized for DevOps Dev Ops Team Tools Services Data Analysis Process
- 60. Azure Web App • Hosting Web Sites • Hosting Web API-s • Hosting Background Workers– Web Jobs • Hybrid Scenarios – Express Route
- 61. DevOps • Built-in staging • Roll-back – Blue / Green • Testing -in-production • Auto Scaling based on custom rules • Performance testing
- 62. Deployment - Auto Swap • Zero Downtime, Zero Cold Start
- 63. Automated Deployment Workflow Internet Traffic Load Balancer Production Slot Staging Slot Application Insights Error Tracker Function Cleanup Function Error Checker Function Error Table PowerShell, Build Server Step Change Traffic Routing Deploy Source Code Swap
- 64. Demo
- 65. Technical Debt - Shared Data Context API #1 API #2 API #3 API #4 API #5Azure SQL Database, Code First Approach
- 66. Issues – Schema change affects all API #1 API #2 API #3 API #4 API #5 DB Schema Change
- 67. Initial State - Coupling
- 68. Group => Split
- 69. Microservices • Small Autonomous services that work together, modelled around a business domain • Small - 2 weeks to rewrite / few hundred lines of code • Independently scalable and deployable
- 70. API Relations API #3 API #8 API #1 API #4 API #5 API #2 API #7 API #6 API #9 6 Callers 4 Callers 2 Callers
- 72. Microservices Characteristics • Very simple services • Use best tool for the job • Loosely coupled • Independent teams • Great for continuous delivery
- 73. Bounded Context
- 74. Microservices Principles • Model around business domain • Hide Implementation (database) • Automation • Chunky, not Chatty Communication
- 75. Microservices Principles • Dumb pipes, smart endpoints • Deploy independently • Highly Observable • Isolate Failure • Chaos Monkey
- 76. Microservices Challenges • Significant Operations Overhead • DevOps skills required • Implicit Interfaces
- 77. Microservices Challenges • Distributed Computing Complexity • Duplication of Effort • Testability Challenges
- 78. Azure Service Fabric • Generally Available • Powers key Azure Services • Great tooling • Pay only for the infrastructure
- 80. Azure Service Fabric • Docker support coming • Azure Stack support coming • Preview for Linux
- 81. Models • Reliable Services: – Stateless – Stateful (reliable collections) • Reliable Actors : – Independent objects (“actors”) – Service Fabric takes care: • Deployment • Scaling • Communication across actors
- 82. DevOps • Any framework supported • Rolling upgrades • Automatic rollback • Application always in consistent state • Fully scriptable -> easy to integrate with CI / CD
- 83. Demo
- 84. Data – The Greatest Asset Dev Ops Team Tools Services Data Analysis Process
- 86. Analysis - Identify Trends=> Improve Dev Ops Team Tools Services Data Analysis Process
- 87. Learn from Errors • Incidents Weekly Review Meeting • Pattern identification • Architecting solutions • Share lessons learnt