Recommended
More Related Content
What's hot
What's hot (20)
Similar to Automation is hard and we are doing it wrong
Similar to Automation is hard and we are doing it wrong (20)
More from Johan Abildskov
Recently uploaded
Recently uploaded (20)
Automation is hard and we are doing it wrong
- 2. Automation is Hard! .. and we are doing it wrong Johan Abildskov @randomsort
- 3. Consulting and Training Continuous Delivery and DevOps A little bit about Praqma and Eficode
- 4. What is this DevOps thing? And where should we begin? CULTURE AUTOMATION LEAN MEASURING SHARING
- 5. What is this DevOps thing? And where should we begin? CULTURE AUTOMATION LEAN MEASURING SHARING
- 6. “We need an army of robots!”
- 8. Our organizations are wrong about automation
- 9. Automation is a luxury!
- 10. It is permission to play
- 13. It is not something we do when the rest is done Automation is not something that we sprinkle on afterwards
- 14. he customer doesn’t want automation!
- 15. 1. Limited software skills in their senior leadership 2. Ambidexterity: the capability to deliver on today’s challenges while also preparing for the future needs of the company 3. Leaders believe that digitalization is an R&D problem 4. Justify their lack of initiative by referring to the lack of desire for change from their most valuable customers
- 16. Automation is not a luxury!
- 19. But our task really is ..
- 22. Architecture Software Architecture is not based on blueprints!
- 23. Automating simple things is simple. Automating complex things might be impossible*. - Me * While retaining sanity
- 24. We, the engineers, are wrong about automation
- 25. Automation is also Software So we should treat it as such
- 29. What is the total cost of ownership of our tools?
- 30. N.I.H Syndrome Not invented here
- 31. We do automation initiatives wrong!
- 34. Manual work vs Automation Amount of Manual work SDO Performance Low Medium High Deployment automation helps you go from low to medium This causes lots more manual testing and change approvals as you release more often Increased test and change approval automation enables the move to high performance Elite
- 35. So what did we try to achieve?
- 36. We must be 50% more productive by 2022 Than what?
- 37. What are we trying to do? And how can we measure it?
- 38. Transactions handled by automation Number of transactions handled by automation
- 39. Lead time on new environments Or time spent setting up testing environments Lead time on new environments
- 40. Regressions errors found in UAT Regressions found in Exploratory Testing
- 42. This forces us to reflect on priorities and actual goals
- 43. Setting goals ● Current Condition ● Target condition ● Deadline ● Action Plan
- 44. There is no silver bullet
- 45. There is nothing so useless as doing efficiently that which should not be done at all. - Peter Drucker
- 46. The silver bullet https://github.com/RandomSort/awesomation/blob/master/silver-bullet.sh $ ./silver-bullet.sh Eliminating wasteful step… $ cat silver-bullet.sh echo “Eliminating wasteful step…” $
- 47. Automation is not a luxury! Thank you for your time!
Editor's Notes
- If we are stuck at the bottom of the stairs of horrible software performance and we want to move towards the promised DevOps nirvana - our first step is very important. For many the first step is automation, and that makes it very important how we approach automation as a part of the beginning of our DevOps Journey.
- How can you expect to be called professional software developers. Tell story of RM. What are some reasonable requirements of providers of custom software?
- State of DevOps 2019
- And this is happening on free software - motivated individuals, as has been mentioned before in the agile manifest. Imagine what you could do with motivated individuals that are handed a budget!
- And so what? The customer doesn’t want TDD, the customer doesn’t want. I would like to claim that we as a business would like to
- Improvement Inflation. It is all well and good that you are better than you were last year. How much are you improving compared to your competition? It doesn’t matter how high you’ve got, it matters how fast you got there!
- An obvious upgrade
- Do you always have something shippable? If you are not able to ship, any feature that you create no matter how magically will never reach the user! This is also the promise of continuous delivery, making deployment a business decision
- Testing is a key capability and a prime target for automation. But do we have the readouts we need, can we connect to the board as needed? Otherwise testing will be complex, testing will be hard
- We can change software architecture. In many cases we can do so incrementially. It is folly to believe that Software architecture doesn’t change, just because it happen to be live.
- And we can solve a lot with instrumentation
- Why is it that we strive to be our best selves, apply scientific thinking, but the instant we smell the opportunity to build a tiny bash script all of our discipline, all of our best practices and hack away. With great pleasure I might add.
- Dancing bear
- They call it the J curve but I’ve drawn it the other way up If we look at the numbers in a slightly different way (y axis changed) Medium performers are doing more manual work than low performers High performers do less than both
- We want to be fifty percent more productive?
- Replacing the API gateway -> measure hits on new vs old gateway
- This is also a way to connect developers to business value. Mention TP Priors, did not know how they impacted business
- Or time spent setting up new environments. Story about testers spending an hour each morning setting up the testing environment
- Could also be errors reported by users
- There is a bunch of this stolen from OKRs
- If you feel like this sounds an awful lot like toyota kata, the ooda loop, scientific thinking, you are right. Names doesn’t matter - different frameworks doesn’t matter. Value from practices matter!