4. Who Am I Anyways?
@FawzyManaa
• Lead DevOps technology lab at Deloitte
• Been focused on DevOps transformation and enablement in large enterprises
• Strong believer in ‘Agile DevOps’ = Fail, Learn, Repeat
• Done many talks at DevOpsDays conferences in North America
5. Why Did DevOps Become a Thing?
@FawzyManaa
• Tension between development and operations teams
• Frustrated IT people not happy at work
• Lean & Agile were already doing wonders to streamline relationship between
business and IT development
• Term was coined in late 2000’s to denote bringing Dev and Ops together
• Movement took off across various community conferences and on Twitter #DevOps
6. How Did It Evolve?
@FawzyManaa
C A L M SCulture Automation Lean Measurement Sharing
These pillars emerged as the necessary foundations for DevOps success:
Happier, more productive people who collaborate more effectively
*Source: Term credited in part to John Willis, Damon Edwards and Jez Humble
7. How Many Organizations Applied It…
@FawzyManaa
C A L M S
Culture Automation Lean Measurement Sharing
Over emphasis on partial automation:
8. How Many Organizations Applied It…
@FawzyManaa
C A L M S
Culture Automation Lean Measurement Sharing
Over emphasis on partial automation:
Automating bad process without changing key behaviours, while not understanding impact
nor telling anyone what was done
9. How Many Organizations Applied It…
@FawzyManaa
C A L M S
Culture Automation Lean Measurement Sharing
Over emphasis on partial automation:
Overwhelmed, less productive people who are more silo-ed and extra protectionist
11. Over-Industrialization of DevOps
@FawzyManaa
While the DevOps movement inspired many companies (especially startups) to adopt a lean
mindset and build with DevOps mindset on green field, many fell in the trap of rapid,
unplanned over-industrialization.
12. Why Do You Need to Know This?
@FawzyManaa
You may already be experiencing
symptoms of rapid, unplanned
DevOps over-industrialization
You’re on the verge of experiencing
these symptoms as you grow and
scale DevOps in your org
Enterprise? Startup?
14. We Need to Provide Some Guarantees
@FawzyManaa
15. First, Don’t Add More Siloes
@FawzyManaa
I want
change
I want
stability
Old Day Problems
OpsDev
I want
change
We’re too cool
for either
Nowadays…
DevOpsDev
I want
stability
Ops
16. DevOps Without a DevOps Silo
@FawzyManaa
Here are some suggested topologies:
Development
Operations Operations
Development
Development
Operations
CoE
Temporary body of evangelists with
expiry date to help enable a new
way of working across the org
Re-structuring to product-based
teams that don’t just develop
and test but also operate
Orchestrating collaboration
between existing development
and operations teams
17. Second, Provide Real Access
@FawzyManaa
We need to automate
deployment for a new
application Okay tell me more
and I’ll create a
Jenkins job for you
DEV OPS DEV OPS
Will you give me
write access?
No, that would
violate segregation
of duties
18. We Need to See the Same Stuff
@FawzyManaa
Plus, that will probably help you solve for this too…
19. Access Laundry List
@FawzyManaa
Failed Builds
Test/ Quality Reports
Console Logs
Application & Server Logs
Monitoring Tools
Performance Dashboards
Planned Outages
…
Commit Code
Create and Approve PRs
Merge/ Push Code*
Trigger & Manage Builds & Deployments
Create & Trigger Tests
Tailor (Write) Pipeline for Application
Co-Manage Infrastructure
…
Access to Know Access to Do
*Dependent on your code review/ approval process
21. Enabling True Collaboration
@FawzyManaa
The following are some techniques that can be used:
Coffee Chats &
Retrospectives
Pull Requests Guidelines, Not Rules
For formal and
informal feedback and
knowledge sharing
To allow dev to
physically contribute to
ops and vice-versa
To enable flexibility
that allows for natural
growth & evolution
22. So What was DevOps Supposed to
Do Again?
@FawzyManaa
23. High Performance Teams
@FawzyManaa
Throughput Stability
Change Lead Time
Time elapsed from code commit to code deploy
Mean Time to Recover
Time elapsed from service outage to service restoration
Change Failure Rate
Ratio of no. of failures in production to no. of changes
to production
Deployment Frequency
No. of deployments per unit of time
Generally accepted metrics to measure successful DevOps adoption:
*Source: High Performance teams metrics via State of DevOps Report published by DORA
27. @FawzyManaa
noun
DevOps First Amendment
Refers to set of provisions that must be guaranteed in any DevOps
approach or journey regardless of contextual difference, target
state, or technology stack.
28. Have more to add?
Share your #DevOpsGuarantee
Find me @FawzyManaa
29. Look forward to connecting with you!
fmanaa@deloitte.ca
Fawzy Manaa
Twitter: @FawzyManaa
LinkedIn: /fawzymanaa
+1 416-525-1813
Get in touch.