Extreme DevOps in Fintech
- 1. Extreme DevOps in FinTech Angelo Agatino Nicolosi Lead Software Architect, Danske Bank
- 2. 11 About Angelo Agatino Nicolosi Lead Software Architect Digital Customer Experience Personal Banking IT Danske Bank Digital Customer Experience Around 180 employees 4 Locations - 3 Countries 10 customer facing applications 60K logons per hour 6K transfers per hour 2.7 Millions customers
- 3. 22 If DevOps was a sport this is how it would look like in a Start-up
- 5. 44 Enterprises must be agile The ability of an organization to sense environmental change and respond efficiently and effectively to that change. “ ” The ability of an organization to renew itself, adapt, change quickly, and succeed in a rapidly changing, ambiguous, turbulent environment “ ”
- 6. 55 The perfect project – Den nye mobilbank Build a new mobile bank completely focused around people’s financial needs which offers outstanding customer experience
- 7. 66 Outstanding customer experience Den nye mobilbank Agility High- Availability Quality
- 8. 77 Stop trying to predict what the business and technology will look like in the future Refocus resources and efforts in building a future-proof architecture.
- 9. 88 Stop trying to predict what the business and technology will look like in the future Refocus resources and efforts in building a future-proof architecture.
- 10. 99 Microservices! Yes! … where do we start? https://martinfowler.com/ ?
- 12. 1111 Never settle! To the Public Cloud and Beyond! Timeline Cloud Talks 2015 01.2016 Experiments on Public Cloud
- 13. 1212 Cloud computing is a risky business. Before anything else think on how to get out if things start to get ugly… Important Stuff #0: Define your exit strategy Business Logic Wiring 80% 20%
- 14. 1313 Enable your legacy to be reusable in order to turn it into competitive advantage instead of rewriting everything from scratch. We already had core banking systems that served us well for decades. Our existing systems Mainframe Channel A Channel B … Channel N Decentral Applications
- 15. 1414 Enable your legacy to be reusable in order to turn it into competitive advantage instead of rewriting everything from scratch. Important Stuff #1: Focus on what is most important. Mainframe (105) Channel A Channel B … Channel N Decentral Applications (102) TECHNICAL DEBT We already had core banking systems that served us well for decades. We had to find where was the most painful part we had to attack first…
- 16. 1515 It happened! Lassie come home! Timeline Cloud Talks Private Cloud is the only cloud 2015 03.201601.2016 Experiments on Public Cloud
- 17. 1616 Important Stuff #2: Keep calm and carry on You know you are right…
- 18. 1717 It happened! Lassie come home! Timeline Cloud Talks Private Cloud is the only cloud 2015 01.2016 Experiments on Public Cloud First PoC in production 03.2016
- 19. 1818 Ensure you can deliver. Always. Important Stuff #3: Always have a Plan B Mainframe (105) Channel A Channel B … Channel N Decentral Applications (102)
- 20. 1919 It’s alive! We are in production! Timeline Cloud Talks Private Cloud is the only cloud 2015 03.201601.2016 Experiments on Public Cloud First PoC in production 06.2016 Microservices in production
- 21. 2020 APIGateway Important Stuff #4: Get to production ASAP. APIGateway MS MS MS
- 22. 2121 APIGateway Ladies and gentlemen, probably the most-expensive proxy in the world! Important Stuff #4: Get to production ASAP. APIGateway Nisba MS MS MS Nada Nix
- 23. 2222 APIGateway Important Stuff #4: Get to production ASAP. APIGateway MS MS MS BAD #1: No containers!
- 24. 2323 APIGateway Important Stuff #4: Get to production ASAP. APIGateway MS MS MS BAD #2: We bought in a solution for a not well understood problem!
- 25. 2424 APIGateway Simplify the setup. Identify and remove bottlenecks. Important Stuff #4: Get to production ASAP. APIGateway MS MS MS
- 26. 2525 APIGateway Start your data adventure easy… what about a caching layer at first? Important Stuff #4: Get to production ASAP. APIGateway MS MS MS Caching
- 27. 2626 Oh yeah! Timeline Cloud Talks Private Cloud is the only cloud 2015 03.201601.2016 Experiments on Public Cloud First PoC in production 06.2016 07.2016 Microservices in production First release
- 28. 2727 Is that it? Are we done?
- 29. 2828 APIGateway Get the granularity right by moving business logic Important Stuff #5: Keep improving! Continuously (Evolutionary!). APIGateway MS MS MS MS MSMS Caching
- 30. 2929 APIGateway Migrate the data when you are sure you are ready and embrace eventual consistency (if feasible). Important Stuff #5: Keep improving! Continuously (Evolutionary!). APIGateway MS MS MS MS MSMS
- 31. 3030 1. Define your exit strategy. 2. Focus on what is most important. 3. Keep calm and carry on. 4. Always have a Plan B. 5. Get to production ASAP. 6. Keep improving! Continuosly. Important stuff
Editor's Notes
- #8: Quality built-in Low latency Minimal amount of bugs in production High-Availability Scalable Distributed Agile Easy to fix Easy to release new functionalities
- #9: In order to achieve the agility we missed we need to embrace evolutionary architectures. Examples? Microservices.
- #10: In order to achieve the agility we missed we need to embrace evolutionary architectures. Examples? Microservices.
- #12: The current situation was that cloud talks were being performed in our enterprise but there was a lot of very high level discussion on Enterprise Architecture, and how this change would impact the enterprise as a whole.
- #13: We have never been good at waiting, and we believed that a different approach was needed. A more practical approach, bottom-up, in order to understand how the wonders of the cloud would meet the expectations from our developers and technical architects, and how would all this fit with the specific requirements a bank needs to comply to. We started to play on public clouds…
- #14: However, we prepared for the worse. An exit strategy was needed in order to avoid wasting the time invested. Moreover, we have been very careful in order to void vendor lock-ins or, if avoidance was not practical or feasible, then we will document and try to reduce the blast radius due to recoding the entire thing against a different technology.
- #15: Reuse what you have always, in order to reduce time to market. Create abstraction layers wrapping legacy in order to free your new applications from the limitations of the legacy.
- #16: Moreover, ensure you have identified where you technical debt actually lies in. Sometimes it is not so obvious… but sometimes to remove it is easier than what you have expected
- #17: It happened! But we did not suffered much…
- #18: You cannot have the cloud now… but you know deep down that sooner or later we will have to embrace it… it is just a matter of time.
- #19: Indeed we just maintained the milestone for the first PoC, by just decoupling frontend and backend development. We made it possible to reuse the existing infrastructure for the frontend project and they could release their PoC without being impacted by the backend migration.
- #20: What did we learn? We didn’t want to go with a WebApp, we needed a Native app to get to the great customer experience we were looking for. Thanks to this approach we did not have to wait for the infrastructure to get production ready.