FINANCIAL TIMES'S JOURNEY TO CONTAINER BASED MICROSERVICES

How we run our applications is always evolving - Monoliths to microservices, migration to the cloud, containerisation, improved orchestration. Containers have become synonymous with microservices but they present a new set of challenges...

In this talk I will share some of our experiences and rationale behind early adoption of containerisation using an in-house written orchestration mechanism through to migrating to Kubernetes.

In this talk I will discuss:

Our motivation for moving services into containers
The pain that went along with very early adoption
Which development approaches and practices help or hinder containerising a service
How our development process changed when we started running in containers
Best practices and lessons learnt around migrations
3 years on: what do we get out of containers now and was the journey worth it?

Transcript

1. Our Journey to Containerised Microservices @nickywrightson nicky.wrightson@ft.com By Nicky Wrightson
2. @nickywrightson Nicky Wrightson Principal Engineer & Tech Lead @ FT
3. https://www.ft.com The news business has changed a lot in the last 130 years
4. How can we do more for less money?
10. https://www.flickr.com/photos/bluesquarething/5165177846 Containers can fix this
11. Over 200 services … In two regions … All containerised …
12. 2. A meandering journey 3. Benefits to us now? 1. Why we chose containers 4. What’s next?
13. 1. Why we chose containers
15. https://six-degrees.ft.com/
16. What is a news website without content?
17. 5+ weeks dev cycle
18. https://twitter.com/architectclippy/status/570025079825764352?s=09
19. Over 200 services … In two regions … ~550 VMs
20. 3 years ago we started to architect for containers
22. We could reduce costs by ~ 80% using containers
23. What about serverless?
24. 2. A meandering journey 1. Why we chose containers
25. Content in Containers == CoCo
26. Containerise every service
27. Monitoring
28. Change database and programming language at the same time!!
29. It was not all plain sailing
30. Stateful services do not play nice
31. “We need a bigger boat” Richard Dreyfus in Jaws (1975) Image: http://goo.gl/4r26gS
32. We kept losing our live clusters
35. Monitor and react at the right levels
37. React quickly to security patching
38. Deployment needs to be automated, consistent, repeatable and have the ability to be rolled back
40. Massive shift in operational thinking
41. If it is hard do it more often If it is really hard automate it!
43. https://coreos.com/blog/migrating-from-fleet-to-kubernetes.html
45. How to iterate on 200 services at the same time?
47. “Stateful services are painful to containerise” Nicky Wrightson January 2016
48. 2. A meandering journey 1. Why we chose containers 3. Benefits to us now?
49. Containers in general can benefit EVERYONE
51. Operationally supportable
52. Kubernetes will fix that ….
53. Deployments are more graceful
54. 2. A meandering journey 1. Why we chose containers 3. Benefits to us now? 4. What’s next?
55. Enhance our initial approach
56. Get someone else to do the heavy lifting
57. One cluster to rule them all …
58. Federation
59. In Conclusion
61. @nickywrightson nicky.wrightson@ft.com Thanks