In case the way I word it helps, containers are a very similar concept to FreeBSD jails or Solaris Zones - so once you get what a container is you may appreciate some of the power that already existed in those systems. It's essentially a partition of the system: picture a kernel, but an additional instance of the network stack for each container, a certain mount-point that becomes the "root" file system for all actions inside that container, etc.
In the end you get very similar functionality to a VM, but because it's just partitions of a kernel it's much more integrated than some third-party hypervisor. This lets you do things like inspect processes that are inside your containers as though they were native processes, etc...
In the end you get very similar functionality to a VM, but because it's just partitions of a kernel it's much more integrated than some third-party hypervisor. This lets you do things like inspect processes that are inside your containers as though they were native processes, etc...