Notes from a talk at LinuxCon/ContainerCon Europe 2016.
The main developer of Traefik presented the software as a talk. He worked at Thales, Airbus, Lyra and Zenika. Then based on work for Traefik founded his own company.
Why do we need another reverse proxy? We already have nginx, haproxy, and others that are already very mature! Mainly because none of these are integrated well with container orchestration frameworks and always require custom plumbing.
He wanted a reverse proxy that integrates with:
- Engine: Docker, rkt
- Orchestration: Swarm, Kube, Mesos, Rancher
- Service Discovery: consul, etcd
Traefik features circuit breakers/rate limiting, weighted round-robin, disaster risk reduction, hot configuration reload, http2, websockets, basic auth for access control. The project boasts 66 contributors. The HTTP stack of the Go standard library is used. Basic metrics are exposed via an API.
Cluster mode, swarm mode, session affinity, better basic auth, generic mesos integration (not only marathon), non-http protocols