DevOps for Speed and Agility - DevOpsTO May 2014
•Download as PPTX, PDF•
0 likes•643 views
How to increase agility and speed in your organization by applying DevOps principles. How to go faster and be more flexible to the needs of your business. Easy steps to reduce bottlenecks, release faster and with more confidence. Improve visibility Measure state Minimize constraints Improve collaboration Enable action Practice change Focus Automate
Recommended
Recommended
More Related Content
What's hot
What's hot (20)
Viewers also liked
Viewers also liked (18)
Similar to DevOps for Speed and Agility - DevOpsTO May 2014
Similar to DevOps for Speed and Agility - DevOpsTO May 2014 (20)
Recently uploaded
Recently uploaded (20)
DevOps for Speed and Agility - DevOpsTO May 2014
- 1. DevOps for Speed and Agility software delivery for fun and profit
- 2. Steve Pereira • DevOps / Delivery consultant in Toronto • <3 startups, feel the pain of enterprise • stevepereira.ca
- 3. To suffer the penalty of too much haste, which is too little speed. - Plato not
- 4. Quality > Speed The catch: The name of the game is to maximize quality and allow confidence and trust to improve speed. Here’s how you get both:
- 5. • Communication • Visibility • Measurement • Analysis • Empowerment • Change • Focus • Automation The Ingredients:
- 6. • Hypothesize • Measure • Validate • Empower • Iterate • Share • Automate • Celebrate The recipe:
- 7. Communicate • Talk about delivery, issues and improvement • Sell ideas, everyone who buys in is a stakeholder and a customer • Talk to product teams about the cost of technical debt • backlog, bugs, lack of documentation • Chat - nothing beats realtime • Standup rooms, team rooms, team standup rooms • History/async helps - it’s self-documenting and remote-friendly • Regular pairing, code review
- 8. Show and Tell • Visibility for simpler and more passive communication • Reduce communication time • Reduce decision time • Reduce investigation time • Start with Post-its/Whiteboard, iterate • Notifications where necessary - alerts vs checking • Correlate data to create meaning
- 9. Measure • Discover limiting constraints (bottlenecks), friction, lag, waste • Start with the basics: How long from dev to prod? • Bugs per release / LOC per release / cyclomatic complexity • Onboarding a dev takes a week • Adding a server takes 3 weeks • 5 days to start a new project • Measure by hand if you have to at first • Find a baseline to progress from
- 10. Minimize Constraints • Remove or minimize bottlenecks once discovered • Hypothesize and validate • Documentation / Tools / Automation • Data and power in the hands of whoever needs it • Ask data questions of data, not people • People are often a constraint • Empower people - Build Trust • Everyone's job is enabling the business
- 11. Practice Change • Make change minimal and frequent • Variables vs hard code • Separate code and config • Break the monolith down • Avoid batch changes, study your use case • Practice deployments / code review / retrospectives • Fire drills!: Staff member leaves / datacentre down
- 12. Focus • Remove disruption to allow for engagement • Clean alerts / Define severity / Scrutinize every escalation / Parse logs • Don't just backlog, icebox - if it’s important you won’t forget about it • Define roles - proper governance allows for action - rotate occasionally • Let your talent work - provide empowerment and time • Standups are important, but try doing them through async chat • Measure a need and hire help or dedicate resources
- 13. Automate • Do you really need manual QA? • Analyze holistically, improve incrementally • Focus on the pipeline • Promoted builds • Config management • Never touch prod • Notifications
- 14. Celebrate! • Build on your momentum and progress by reflection and sharing • Improvement is awesome! Faster iteration means more to celebrate • Failure hurts less the more you do it and the less it costs • Share with the entire organization, wins help everyone • Give kudos to your champions, testers and early adopters - they’re your best customers
- 15. The Advanced Class • Tools can help once you’re off to the races • Jenkins + plugins is a powerhouse: • Build metrics - stats on all builds • Plot - graph progress • Join - breakup jobs and aggregate results • SLOCCount - LOC counts • Violations - static analysis • HTML publisher - show it all off
- 16. Speed provides the one genuinely modern pleasure. - Aldous Huxley
- 17. Start now • Pick a small, greenfield project • Fail like a pro • Write about it • Share • Revise your baseline and repeat
- 18. Questions? Sites: codeascraft.com martinfowler.com kitchensoap.com planetdevops.net monitorama.com devopsdays.org Netflix/Twitter/Linkedin Eng Books: Continuous Delivery Release It! Building a DevOps Culture Driving Technical Change The Mythical Man Month The Phoenix Project Team Geek @steveElsewhere
Editor's Notes
- Improve visibility Measure state Minimize constraints Improve collaboration Enable action Practice change Automate Focus
- The groundwork for going fast Nothing can be fast without being properly engineered Starts with a commitment to quality Incremental improvement
- Quality over speed The name of the game is to maximize quality and allow confidence and trust to improve speed
- Communication Visibility Measurement Analysis Empowerment Change Automation Focus
- Hypothesize Measure Validate Empower Iterate Share Automate Profit
- Objective: Share an understanding or vision - the idea that things can go faster Talk about delivery Make it known it’s a concern of yours Find out who’s interested Find allys Once you get traction, you’ll have to deal with improving communication and communicating issues - next section Technical debt: - 20% new features - 80% existing features You will need feedback and buy in
- Objective: Visibility for simpler and more passive communication reduce communication time reduce decision time reduce investigation time dashing / geckoboard / new relic / kibana / sentry / airbrake jenkins radiator / dashboard / project statistics Correlate - deploys with server performance and user experience Push vs pull - zapier / deadmanssnitch
- Objective: Discover limiting constraints, lag, waste Start with the basics: Change lead time Bugs per release / LOC per release / cyclomatic complexity Find friction/bottlenecks/constraints Scalability, ease of onboarding, context switching Adding a server takes 3 weeks Onboarding a new dev takes a week Switching projects takes a day Stopwatch, time command, subtracting timestamps People are often a constraint
- Objective: repeatability and consistency are key to improvement selenium / browserstack browsershots / wraith jenkins / circle / travis
- Objective: Build on momentum and progress by reflection and sharing
- Jenkins plugins
- The groundwork for going fast Nothing can be fast without being properly engineered Starts with a commitment to quality and trust Incremental improvement