Agile and DevOps both aim to break down barriers between development, testing, and operations teams. Agile focuses on small incremental updates and open communication, while DevOps emphasizes continuous integration, deployment, monitoring metrics and automation. Both approaches are meant to help teams work collaboratively across functions, but should not involve having teams take on responsibilities outside their normal roles. The key principles of increased ownership, automation, and metrics-driven design are often only partially implemented.

1. Agile:
An Overview
Narayanan Subramaniam
LinkedIn: http://www.linkedin.com/in/cnsubramaniam
@credits for images thanks to various internet sources

2. (1)
Development
(2) Validation
(3)
Deployment
(4)
Production
(5)
Support

3. @copyright: Narayanan Subramaniam (9341969647)@credit: Boeing SW … https://apps.dtic.mil/dtic/tr/fulltext/u2/a558044.pdf

4. ü Agile attempts to break the barriers between Dev and Test and focuses on small
increments while failing fast with emphasis on openly communicating feature
behaviours, expectations and results
ü Agile is NOT development engineers doing QA full time, or Architecting /
Designing on the fly WITHOUT initial thought/groundwork or reviews - such
misinterpretations usually lead to disasters. Cross-pollination helps a lot but is with
the purpose of creating heightened awareness.
ü The core principles of taking more ownership, automation of test, as key to
survival, are often missed out or partially adopted.
ü It is a different question whether every project is amenable to Agile or not ...
certain projects may not be suited for Agile at the outset.
ü Focus on the principles, not just the structure of teams, meetings etc.

5. DevOps:
An Overview
Narayanan Subramaniam
LinkedIn: http://www.linkedin.com/in/cnsubramaniam
( +91-9341969647)

6. Agenda:
ü DevOps …. is it all about Tools, PagerDuty and the Cloud ?

7. (1)
Development
(2) Validation
(3)
Deployment
(4)
Production
(5)
Support

8. Challenges:
ü How do I fix my bug, and quickly roll it out ?
ü How can I pilot my new feature “carefully” ?

9. Challenges:
ü What are the eco-systems my software is deployed in ?
o Configuration combinations
o Interop combinations
o …….

10. Challenges:
ü How is my software used ?
o What features are used ?
o Who uses it and when/why ?
o What is the UE (User Experience) ?

11. Challenges:
ü Is the software “vulnerable” ?
o How well does it perform ?
o Does it scale as expected ?
o Is the software being “attacked” ?

12. Challenges:
ü How do I know if the design is “future-proof” ?
o Will it run out of capacity soon ?
o How many more users can it support - SLA ?
o How do I plan for “breaking” changes ?

13. (1)
Development
(2) Validation
(3)
Deployment
(4)
Production
(5)
Support

14. ü How do I fix my bug, and quickly roll it out ?
ü How can I pilot my new feature “carefully” ?
o Continuous Integration & Deployment (CI/CD)
Ø Plan Big, Execute Small - Architecture/Design is key
Ø Incremental Code Development & Test – AGILE

15. ü How do I fix my bug, and quickly roll it out ?
ü How can I pilot my new feature “carefully” ?
o Continuous Integration & Deployment (CI/CD)
Ø Preferably one source tree
Ø Automate – builds, installs, tests, configuration

16. ü How can I pilot my new feature “carefully” ?
o Continuous Integration & Deployment (CI/CD)
Ø Highly Available – Microservices architecture/design
Ø Fail-fast and Fail-forward

17. ü How can I pilot my new feature “carefully” ?
o Continuous Integration & Deployment (CI/CD)
Ø Feature Toggles – Canary Releases

18. ü What are the eco-systems my software is deployed in ?
o Multiple Data-Center, Regions/Geographies ?
o Self versus Partner Hosted ?

19. ü How is my software used ?
ü Is the software “vulnerable” ?
ü How do I know if the design is “future-proof” ?
Alarms, Events
Raw Counters, Thresholding
Debug, Trace Logs, Backtrace
Health Checks
Event/Alert
Management
Systems
Performance
Management
Systems
Log Analysis
Systems
Management
Platform
Metrics & Analytics
System
Application
Software
Composite Metrics

21. ü Metrics are Ever Changing !!
o Leverage CI/CD not for features alone, but for metrics

22. ü Automated Orchestration based on Metrics and Events

23. ü DataScience and Machine Learning for Scale

24. ü Regulation
o Data Export: Data crossing regions, countries
o Data At Rest: Data retention policy, access control
o Legal Intercept
ü PII (Personally Identifiable Information):
o Obfuscation, RBAC to Operations/Admins to reverse-map PII data

25. ü SRE as a Gatekeeper to SLA’s

26. ü Two Backlogs !!
o Feature Backlog – Product Management driven
o DevOps Backlog – Metrics and Automation driven
o Team Rotation helps

27. Takeaway:
ü DevOps:
o Mindset – E2E
o Agile execution
o Automation
o Metrics, ML
o Regulation, PII
o Multiple Backlogs
ü Focus on the Principles, Tools are incidental
ü SME, Architects, Product Management remain key stakeholders
ü DevOps Principles applicable beyond Cloud based products

28. ü DevOps attempts break the barrier between Dev, Test and Operations with ever
increasing emphasis on automation not just of test but of operations as a whole.
ü Metrics based design is the key paradigm for DevOps. It usually builds on the
goodness of Agile TDD.
ü Agile is NOT about development engineers doing Operations or Operations folk
writing feature code. Cross-pollination helps a lot but is with the purpose of
creating heightened awareness. Such misinterpretations usually lead to disasters.
ü The core principles of taking more ownership, metrics based design to quantify
SW quality/usability, automation of operations and support, as key to survival, are
often missed out or partially adopted
ü Focus on the principles, “culture, measure and share”, and not just on “automate”

29. THANK YOU !