Scaling Infrastructure Without Scaling Ops

If everything around you is exploding by growth, how can your ops teams keep up with that? We'll explain how we managed to do it succesfully and how we are going to use the cloud to scale infinitely.

Transcript

1. Frank van Boven & Roelof Reitsma | 09-05-2017
5. Scaling infrastructure at Coolblue Content • Scaling steps of Coolblue through the years • Non-technical talk / Warstories • From nothing to the Cloud • Babysteps • Success and failures along the way
8. Business plan
10. Can we make a billion?
12. 2003: scaling everything Fresh start • Not-scalable homebrew software • Not enough people • Next step: File new project
15. Big Bang release
19. 2010: Scaling issues No more space • Software quality and quantity improved • Shared hosting could not extend • Server room under kitchen has not enough power • Next step: buy rack in data centre
24. M1 Database
26. M1 Database Database Database Database Database Database Database Database Database
27. M1 Database Database Database Database Database Database Database Database Database
28. Patch management...
29. 2013: Scaling issues Losing control • 50+ servers in data centre • Operations team managing infrastructure • Manually created, undocumented and inconsistent configuration • Lack of proper patch management • Next step: configuration management
31. Configuration management
32. M1 Database Database Database Database Database Database Database Database Database
36. 2015: Scaling issues Ops team bottleneck • 75+ servers in configuration management • Time to deliver under stress • Lot of teams dependent on ops team • Next step: use the force!
39. Puppet: the good, the bad & the ugly
40. E-commerce architecture - 2016
42. M1 Database Database Database Database Database Database Database Database Database Database Database Database Database
43. Scaling for Christmas 2016 • Find appropriate hardware • Get offers and approval • Buy hardware • Install hardware • Install and configure software 1 day 2 months Ordering new hardware
44. 2016: Scaling issues Peak scaling • Easy to roll out new systems • Complex infrastructure environment • Big difference in peak and low usage • Next step: go to the cloud
49. Transparency
50. Managed Services
54. Can we make a billion?
59. VS