Upload
others
View
5
Download
0
Embed Size (px)
Citation preview
Scaling a conventional on-premise CI infrastructure to
the cloudMikhail Advani
Sebastian CarlierMendix B.V.
Who are we?● Mikhail Advani - Site Reliability Engineer at Mendix
○ Twitter - @mikhail_advani○ Website - https://mikhailadvani.github.io
● Sebastian Carlier - “Devops” Engineer at Mendix○ Twitter - @c4rlier
What were we shipping?
...and how?
Symptoms of the problem● Tight coupling between build jobs and hardware● Periodic housekeeping on servers● Difficulty in upgrading infrastructure● Long feedback loop for builds
Source: https://xkcd.com/303/
Design Goals● Fresh build agents with every run● Immutable infrastructure● Scalable infrastructure● Infrastructure usage overview
Migration Strategy
Infrastructure as Code● Infrastructure
Infrastructure as Code● Infrastructure
○ Amazon Web Services○ Terraform○ Kubernetes via Kops○ RDS
Infrastructure as Code● Infrastructure
○ Amazon Web Services○ Terraform○ Kubernetes via Kops○ RDS
● Configuration management
Infrastructure as Code● Infrastructure
○ Amazon Web Services○ Terraform○ Kubernetes via Kops○ RDS
● Configuration management ○ Packer with Ansible○ Docker
Code Architecture
How do we secure this infrastructure?
What did dynamic scaling ever do for us?
Resource utilization
What really worked for us?
If we were to do this again, what would we change?
Thanks
https://xkcd.com/1629/