Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.
Why it’s a good idea 
to deploy 10 times per day 
2014-10-08 
Joakim Lindbom 
Principal | Enterprise Architect
Now, what’s this?
Speed Kills!
90 
80 
70 
60 
50 
40 
30 
20 
10 
0 
100 
Lack of speed kills! 
2001 2003 2005 2007 2009 2011 2013 2015 
Best & Beautifu...
Zero-Day forever 
Your time to react on errors will approach ZERO 
Will become reality with the growing mobile market and ...
Zero-Day forever 
Your time to react on errors will approach ZERO 
Will become reality with the growing mobile market and ...
Termodynamics 
Second main law 
The amount of 
caos increases 
by itself
Design-Build-Run
Design-Build-Run approach 
Reqs 
Specify 
system 
Build SW 
Design 
system 
Write 
code 
Build 
system 
Install 
system 
T...
http://www.capgemini.com/resources/world-quality-report-2013-14
https://www.worldqualityreport.com
Cloud = access to abundance
Cloud = access to abundance 
You can have as many development, test & staging 
environments as you like! 
But you cannot h...
Big IT  Slow IT 
(not slow as in slow food….)
OFF THE PACEHow slow is slow? 
The pace of digital transformation is too slow – unless you’re the CEO. 
12% 
33% 
25% 
Who...
IT Legacy is #1 obstacle for innovation 
Major corporations depend on core systems that 
- Are 15-20 years old 
- 3-4 pers...
IT Legacy is #1 obstacle for innovation 
Major corporations depend on core systems that 
- Are 25-45 years old 
- 1-2 pers...
The ability to innovate 
is strictly coupled with 
the ability to fail fast
Nine observations on IT Complexity 
1. Most IT systems are too complex. 
2. "Best Practices" increase complexity. 
3. Comp...
But what is Simple-IT, then? 
1. Non complex 
2. Small building blocks 
3. Autonomous 
4. They “know nothing” 
5. Service ...
But what is Simple-IT, then? 
Simplified 
Modularised 
Optimised 
Hardened
DevOps is about increasing your 
responsiveness to customers
Devs  New features, fast! 
Ops  Uptime, uptime & uptime
But whatif… 
Devs  Uptime 
Ops  New features, fast!
Or…. 
…should it be one joint team per system/service?
Autonomous system? 
Totally separate parts 
Separate lifecycle! 
Loose coupling paw rihk-titt
Autonomous system? 
Stuff Data 
System 
We used to look at it from a technical boundaries 
perspective
Autonomous system? 
Stuff Data 
Architect Dev Test 
System 
Ops 
But in order to ensure agility, we need to include 
the p...
Autonomous system? 
Stuff Data 
Architect Dev Ops Test 
System 
OpenAPI 
Open 
Data 
Lake 
OpenAPI mindset = not a solutio...
~Autonomous system? 
Release = just a mountain to climb…
Basics, get things in order 
Ad-hoc deployment 
Structured & planned releases 
Major hurdle Major RISK 
Major event
Climbing a Release-mountain, how many people 
experience it…
Industrialisation of IT 
IT development more and more viewed as 
"manufacturing" 
Square boxes, repeat over and over 
Mist...
How good is good? 
When your developers check in code, how 
good does it need to be? 
Compileable? 
No warnings? 
Runnable...
Deploy 10 time per day 
How will that help you?
Deploy often 
Shorten feedback-loop 
Bring back passion – show visible result early 
Allow (small) failures 
Allow experim...
Deploy often 
Basis for Continuous Improvement 
Slow break-down into μServices 
Gradual transition towards smaller buildin...
Continous improvement 
Continuous rebuilding 
Always. Even if “not needed”
Example
Build chain 
Does it 
compile 
? 
Unit 
tests 
OK? 
SW 
quality 
metrics 
Build Deploy 
trunc 
Auto 
Check-in 
Test
Build chain 
Does it 
compile 
? 
Unit 
tests 
OK? 
SW 
quality 
metrics 
Build Deploy 
trunc 
Manual 
Check-in 
Productio...
Tools 
Hudson/Jenkins/Ant – base automation 
Phing – PHP specific manipulation tool 
SonarQube – Software quality, general...
Summing up 
Lack of speed kills 
DevOps – increase responsiveness 
Simplify, make smaller & rebuild – to fight 
complexity...
Contact 
Joakim 
Lindbom 
Principal | Enterprise Architect 
Joakim.Lindbom@capgemini.com 
08-5368 3934 
0708-166404 
twitt...
Image sources, marked as OK to use commercially 
https://upload.wikimedia.org/wikipedia/commons/8/8b/Buck_Mountain_Grand_T...
2014-10 DevOps NFi - Why it's a good idea to deploy 10 times per day v1.0
2014-10 DevOps NFi - Why it's a good idea to deploy 10 times per day v1.0
Upcoming SlideShare
Loading in …5
×

2014-10 DevOps NFi - Why it's a good idea to deploy 10 times per day v1.0

1,644 views

Published on

Corporations are struggling with overly complex systems and system landscapes. DevOps is presented as one piece of the puzzle to go for much leaner and simpler landscapes - all in order to increase the readiness for change and innovation.
The presentation also discusses the the basic thought error behind organising according to Design-Build-Run, which is the basis for most ICT IM outsourcing.

Published in: Software
  • Be the first to comment

  • Be the first to like this

2014-10 DevOps NFi - Why it's a good idea to deploy 10 times per day v1.0

  1. 1. Why it’s a good idea to deploy 10 times per day 2014-10-08 Joakim Lindbom Principal | Enterprise Architect
  2. 2. Now, what’s this?
  3. 3. Speed Kills!
  4. 4. 90 80 70 60 50 40 30 20 10 0 100 Lack of speed kills! 2001 2003 2005 2007 2009 2011 2013 2015 Best & Beautiful A Inc B Inc C Inc
  5. 5. Zero-Day forever Your time to react on errors will approach ZERO Will become reality with the growing mobile market and IoT.
  6. 6. Zero-Day forever Your time to react on errors will approach ZERO Will become reality with the growing mobile market and IoT.
  7. 7. Termodynamics Second main law The amount of caos increases by itself
  8. 8. Design-Build-Run
  9. 9. Design-Build-Run approach Reqs Specify system Build SW Design system Write code Build system Install system Test Reqs Test Specs Integr. test Test design Unit test System test UAT A bit too waterfallish
  10. 10. http://www.capgemini.com/resources/world-quality-report-2013-14
  11. 11. https://www.worldqualityreport.com
  12. 12. Cloud = access to abundance
  13. 13. Cloud = access to abundance You can have as many development, test & staging environments as you like! But you cannot handle this manually!
  14. 14. Big IT  Slow IT (not slow as in slow food….)
  15. 15. OFF THE PACEHow slow is slow? The pace of digital transformation is too slow – unless you’re the CEO. 12% 33% 25% Who are these 10% 8% 8% 39% guys?!? 3% 2% 3% 3% 22% 36% 42% 43% 46% 12% 15% 22% 23% CEO/ President/ Managing Director C-Level executives and board members Managers Staff Very Fast Fast About right Slow Very Slow MIT Center for Digital Business and Capgemini Consulting
  16. 16. IT Legacy is #1 obstacle for innovation Major corporations depend on core systems that - Are 15-20 years old - 3-4 persons know - Some staff is retired, some will be it within 4-5 years - Technology support is slow - Are somewhat documented - Have few formal test cases, but the staff know how to test Application Landscape Report 2014
  17. 17. IT Legacy is #1 obstacle for innovation Major corporations depend on core systems that - Are 25-45 years old - 1-2 persons know - All staff are retired, or will be it within 2-3 years - Technology support is gone - Are undocumented - Have no/few formal test cases Application Landscape Report 2014
  18. 18. The ability to innovate is strictly coupled with the ability to fail fast
  19. 19. Nine observations on IT Complexity 1. Most IT systems are too complex. 2. "Best Practices" increase complexity. 3. Complex systems cost more to build. 4. Complex systems are harder to deliver. 5. Complex systems are less secure. 6. Complex systems are less reliable. 7. Complex systems are less agile. 8. Complex systems cost more to run. And 9. Existing management approaches ignore complexity.
  20. 20. But what is Simple-IT, then? 1. Non complex 2. Small building blocks 3. Autonomous 4. They “know nothing” 5. Service based 6. Dynamic 7. Like lego bricks 8. Exposes an OpenAPI 10. Joint development and operations Knowing nothing? Image: Daily Mail
  21. 21. But what is Simple-IT, then? Simplified Modularised Optimised Hardened
  22. 22. DevOps is about increasing your responsiveness to customers
  23. 23. Devs  New features, fast! Ops  Uptime, uptime & uptime
  24. 24. But whatif… Devs  Uptime Ops  New features, fast!
  25. 25. Or…. …should it be one joint team per system/service?
  26. 26. Autonomous system? Totally separate parts Separate lifecycle! Loose coupling paw rihk-titt
  27. 27. Autonomous system? Stuff Data System We used to look at it from a technical boundaries perspective
  28. 28. Autonomous system? Stuff Data Architect Dev Test System Ops But in order to ensure agility, we need to include the people in the system definition
  29. 29. Autonomous system? Stuff Data Architect Dev Ops Test System OpenAPI Open Data Lake OpenAPI mindset = not a solution design for a specific purpose/project. Open for Innovation
  30. 30. ~Autonomous system? Release = just a mountain to climb…
  31. 31. Basics, get things in order Ad-hoc deployment Structured & planned releases Major hurdle Major RISK Major event
  32. 32. Climbing a Release-mountain, how many people experience it…
  33. 33. Industrialisation of IT IT development more and more viewed as "manufacturing" Square boxes, repeat over and over Mistake! IT is innovation
  34. 34. How good is good? When your developers check in code, how good does it need to be? Compileable? No warnings? Runnable? Passing tests? Not breaking anything else?
  35. 35. Deploy 10 time per day How will that help you?
  36. 36. Deploy often Shorten feedback-loop Bring back passion – show visible result early Allow (small) failures Allow experimentation Follow Moore’s law Learn by doing But you don’t need to deploy to production 10 times per day…
  37. 37. Deploy often Basis for Continuous Improvement Slow break-down into μServices Gradual transition towards smaller building blocks
  38. 38. Continous improvement Continuous rebuilding Always. Even if “not needed”
  39. 39. Example
  40. 40. Build chain Does it compile ? Unit tests OK? SW quality metrics Build Deploy trunc Auto Check-in Test
  41. 41. Build chain Does it compile ? Unit tests OK? SW quality metrics Build Deploy trunc Manual Check-in Production Dev-test
  42. 42. Tools Hudson/Jenkins/Ant – base automation Phing – PHP specific manipulation tool SonarQube – Software quality, general PHPUnit + LoC, Copy/Paste Detector, Mess Detector, et all http://www.sonarqube.org/ https://github.com/sebastianbergmann
  43. 43. Summing up Lack of speed kills DevOps – increase responsiveness Simplify, make smaller & rebuild – to fight complexity Automate to avoid simple misstakes!
  44. 44. Contact Joakim Lindbom Principal | Enterprise Architect Joakim.Lindbom@capgemini.com 08-5368 3934 0708-166404 twitter: JoakimLindbom http://www.slideshare.net/JoakimLindbom http://www.linkedin.com/in/joakimlindbom
  45. 45. Image sources, marked as OK to use commercially https://upload.wikimedia.org/wikipedia/commons/8/8b/Buck_Mountain_Grand_Teton_NP1.jpg https://upload.wikimedia.org/wikipedia/commons/d/de/Eisklettern_kl_engstligenfall.jpg https://c1.staticflickr.com/1/1/1118807_a751d65ba5_z.jpg?zz=1 https://upload.wikimedia.org/wikipedia/commons/4/4f/Eternal_clock.jpg https://c1.staticflickr.com/9/8062/8189938256_2a683d2334_z.jpg https://upload.wikimedia.org/wikipedia/commons/8/85/Git_branches_example.png https://upload.wikimedia.org/wikipedia/commons/9/99/Highway_at_night_slow_shutter_speed_ph otography_02.jpg https://c2.staticflickr.com/6/5058/5490790304_dc3d7c2b91_z.jpg http://www.dailymail.co.uk/tvshowbiz/article-1176568/Andrew-Sachs-thanks-Jonathan-Ross- Russell-Brand-boosting-career.html https://upload.wikimedia.org/wikipedia/commons/8/8b/CERN_Server.jpg

×