scaling n26 technology through hypergrowth
play

Scaling N26 Technology Through Hypergrowth @folgerfonseca QCon - PowerPoint PPT Presentation

Scaling N26 Technology Through Hypergrowth @folgerfonseca QCon London 03.2020 Building the bank world loves to use A new approach to banking Best digital user A betuer price experience +5M +$1.66Bn Customers Monthly transaction volume


  1. Scaling N26 Technology Through Hypergrowth @folgerfonseca QCon London 03.2020

  2. Building the bank world loves to use

  3. A new approach to banking Best digital user A betuer price experience

  4. +5M +$1.66Bn Customers Monthly transaction volume 26 4 Markets Locations +$680M +1500 Funding Employees

  5. SPACES Organise your finances.

  6. +180 +60 Microservices Cross-functional teams N26 tech 100% +500 In the cloud Deployments per week

  7. but How?

  8. BACKGROUND

  9. 3 Years ago 6 +100K Markets Customers +40M +100 Funding P&T Employees

  10. Our challenges Small team Time critical Relatively young

  11. Our challenges Small team Time critical Relatively young

  12. Experimentation & Knowledge sharing 2 Days GSDD Lightning talks 1x every 6 weeks 1x Week

  13. HYPERGROWTH

  14. Hypergrowth

  15. Our challenges → More User More Markets → → More Tech Hubs More People → → More Security More Stability →

  16. PEOPLE SERVICES RELEASES RELIABILITY

  17. Our Goal Bring in more teams to enable us to get more users and release the in new markets

  18. Company onboarding Agile QA SRE Product Backend Security

  19. Team onboarding

  20. Buddy system

  21. Target operating model Segments Group Group Team Team

  22. Target operating model 3 Different currencies 3 Different banking regulations +10 Payment schemes 4 Different locations

  23. Our Goal Have consistency in technology in order to build up knowledge, enable team mobility

  24. Microservices allows you to write each new service in a different language.

  25. Languages

  26. Decisions & Alignment Tech Radar - Based on Thoughtworks - Safe way to try things - Favor Consistency - Alignment across teams

  27. PEOPLE SERVICES RELEASES RELIABILITY

  28. Problem How to avoid micro-monoliths

  29. IaC Infrastructure as Code

  30. Infrastructure as code Servers Network Setuings

  31. Strong separation → AWS organisations enables very fine-grained segregation of services Scalable → Small number of people-to-server ratio Infra as Code Agile → Supporting change or new services easy Benefits Reduced human error → Automation prevents deviation Source controlled → 4 eyes principle plus audit

  32. Example Remember meltdown and spectre?

  33. Meltdown Spectre

  34. “ Affected billions of systems globally

  35. Vulnerabilities publicly announced

  36. Intel responds with initial statement

  37. Intel announces a fix “for the majority”

  38. Amazon announces updated Kernel

  39. Base image updated and tested

  40. Deployments with new image incrementally tested and rolled out

  41. Result? Done!

  42. “ the speed of a startup and the security of a bank

  43. PEOPLE SERVICES RELEASES RELIABILITY

  44. Continuous Delivery CD

  45. Continuous Delivery Security and Artefact Automated Automated Automated automated End Publish checks checks checks testing Code Review Dev Staging Live Canary Live roll Artefact build and commit Deploy Deploy Deploy out Sign Off

  46. Continuous Delivery - Live Blue Green LB Instance Instance Instance Instance Instance

  47. Continuous Delivery - Live Blue Green LB +1 Instance Instance Instance Instance Instance

  48. Continuous Delivery - Live Blue Green LB +1 +1 +1 +1 +1 Instance Instance Instance Instance Instance

  49. Immutable infrastructure → No snowflakes in our infrastructure Rollbacks are easy → Historical AMIs enable rollback CD Scalable → Benefits Multiple instances enable scaling control Fine-grained control → Implement fine-grained security and compliance controls Customisable → Good machine selection for types of services

  50. PEOPLE SERVICES RELEASES RELIABILITY

  51. Problem While having a high release rate, keep high availability and Minimize risks.

  52. High Availability >500 deployments per week Incidents

  53. “ The best way to avoid incidents is by avoiding changes

  54. Minimize time Delivery Fix Detection Diagnostic

  55. High Availability Blameless Postmortem culture

  56. A written record Preventing actions Actions to What includes mitigate post mortem? The incident impact The root cause

  57. PEOPLE SERVICES RELEASES RELIABILITY

  58. Adjust to your journey conditions Trade off analysis Continuous Delivery Is enable trouble healthy automation Lessons learned Infra-as-code AWS makes it easy to implement Keep a availability in mind Keep a close eye on service objectives

  59. Scaling N26 Technology Through Hypergrowth Thank you! Any questions? @folgerfonseca

Recommend


More recommend