Reducing inertia in organizations is the key to a successful DevOps transition

A presentation at DevOpsDays Riga 2018 in September 2018 in Riga, Latvia by Joep Piscaer

Slide 1

Slide 1

Slide 2

Slide 2

Hello, I’m Joep Piscaer Technical Pathfinder @jpiscaer #DODR2018 #DevOpsDaysRiga

Slide 3

Slide 3

3 concepts • Autonomy • Flow • Simplicity Solve inertia of • Team culture • Finance • Infrastructure

Slide 4

Slide 4

Autonomous & multidisciplinary Teams • Independently develop and release into production • Have all the skills, roles and tools to reach team goal and mission • Don’t hide behind cultural inertia and defense mechanisms • Operate outside comfort zone • Responsible for a business outcome, not a business function

Slide 5

Slide 5

• Align systems with teams • Take ownership of system • Break down dependencies • Loose coupling reduces complexity • Take advantage of Conway’s Law Align tech systems along organizational boundaries

Slide 6

Slide 6

Organize in cherries Product owner Team lead • Building block of the agile organization • Technologically adjacent teams form tribes • Work and solve similar issues together reduce ripple effect to smaller, less complex scale • Teams are small, 5-8 members • Do by-the-book scrum Scrum teams Cell Aligned to a tech domain

Slide 7

Slide 7

• Freedom gives bottom-up choice • No forced usage of central IT resources (infrastructure, software) • Choose your own resources • Different teams choose differently Ownership gives freedom

Slide 8

Slide 8

Part of pipeline Low-friction Infrastructure & Public Cloud Byte-size Pay-asyou-go Selfservice Ready-touse

Slide 9

Slide 9

Use COTS Software: Or develop completely custom In a standard way Buy or Build? Customization is loosely coupled As intended

Slide 10

Slide 10

Infrastructure in the value stream • Cloud and Infra expertise embedded in the value stream • Infra is not the biggest bottleneck in the pipeline anymore • Work on business outcome (speed, value), not ’central IT’ function (cost control) • Unfit infrastructure, wasteful handovers, bureaucratic ticketing system and slow approval gates annihilated

Slide 11

Slide 11

Think small • Single unit flow through the pipeline • Short feedback loop with immediate result • No batch processing of commits • Minimal amount of work-in-progress • No code waiting to go to production • Nothing stuck in the pipeline • No unused artifacts produced • No context switching for developer • No loss of information during handover • But only where it makes sense

Slide 12

Slide 12

Preventing re-work fail small and learn immediately • • • • • • Team is responsible for running code in production All operational aspects, maintenance, roadmap Code is tested thoroughly before release Team does investigation, mitigation and post-mortem Feedback into pipeline to prevent and improve Incentivizes ’first time right’ and quick remediation

Slide 13

Slide 13

Expertise in Chapters • Could call them Pathfinders • Experts in their field • Coaching and Learning as primary output • Team Leads gravitate to soft skills • Pathfinders gravitate to hard skills • Pathfinders lead Guilds and Chapters • Are not HR-responsible • Play a big part in technical overview (‘architecture’) • Lead the bigger initiatives

Slide 14

Slide 14

‘Central IT’ is a decentralized guild • Manage the end-to-end collection of connected simplicities ( ‘architecture’) • Standardize design patterns and cloud consumption across teams (‘operation’) • Safeguard non-functional aspects • • • • • • • Cost optimization (buy as a group) Identity & Access Management Observability & Monitoring Security, compliance Performance Reliability Risk management (lock-in)

Slide 15

Slide 15

Pick one or two to take with you Build autonomous & multidisciplinary Teams in ‘Cherry’ structure That are responsible for a business outcome, not a business function Who’ll align tech systems along organizational boundaries break down dependencies and think small Have infra & cloud expertise embedded in the value stream Who will not use unfit infrastructure, wasteful handovers, bureaucratic ticketing system and slow approval gates Because they are responsible for running code in production And want minimal rework and errors (‘first time right’ and ‘fail small’) ‘Central IT’ is a decentralized guild Pathfinders lead Chapters Which manages architecture as a collection of connected simplicities Help standardize design patterns and cloud consumption across teams And safeguard nonfunctional aspects Tweet me your fairytale story @jpiscaer