Using LXC to scale various workloads requires a proxy service such as HAProxy or Nginx, and the actual service running in containers. Network connectivity and segmentation can be achieved using software-defined networks with OVS and GRE tunnels.
In this chapter, we looked at how to run Apache in simple libvirt-based LXC containers that do not require an entire root filesystem, but rather a minimal set of directories with the shared binaries and libraries from the host OS. We also created an Apache cluster behind a load balancer on multiple servers, and demonstrated a simple yet effective way of scaling it by duplicating an LXC container.
In the next chapter, we'll build on what you've learned so far, show how to monitor and back up LXC, and create an autoscaling service with Jenkins and Sensu.