Herd of Containers - Saâd Dif

BlaBlaCar's infrastructure journey: From Chef to containerization with Docker, CoreOS, and Fleet, and managing high availability with BGR, PostgreSQL, and more.

Key takeaways

BlaBlaCar’s infrastructure is based on containers, using Docker, CoreOS, and Fleet to orchestrate them.
They use BGR (Bidirectional Replication) to replicate data across multiple nodes for high availability.
PostgreSQL is their main database, with 100GB of data and 50 million travelers using the platform.
They used to manage their infrastructure with Chef, but switched to a more container-focused approach.
They use Grafana to create dashboards and visualize metrics, with targets defined for specific points.
The team uses Prometheus and a custom tool called DGR to handle metrics and alerting.
For storage, they use Cassandra, with an emphasis on high availability and scalability.
BDR (Bidirectional Replication) is used for disaster recovery, with replication lag being a main concern.
The team uses Zookeeper for distributed locking and coordination.
They use Elasticsearch for search and filtering, with high availability in mind.
Their containers are configured using systemd and etcd, with annotation used to provide summary information.
The team is constantly improving their infrastructure, with a focus on performance, scalability, and reliability.

Herd of Containers - Saâd Dif

More talks