With a community of over 40 million members, BlaBlaCar has to scale and automate its platform continuously to keep it fast and reliable. Two years ago year, the foundations team at BlaBlaCar has decided to optimize the core infrastructure by going all-in on containers. Now we have been running all our production services with rkt and CoreOS for several months. This talk is about how BlaBlaCar changed the PostgreSQL default topology to make a node expendable in a containerized environment.
Subjects : - Service Discovery - Containers (rkt) - Multi Master Logical Replication (BDR) - Monitoring and alerting with Prometheus - Backup and Recovery