Docker Orchestration
As Docker is called on to work with larger systems, and particularly in the cloud, the complexity of coordinating where containers are running and why increases exponentially. In response, many companies have developed Docker “orchestration” systems to manage this complexity. Because this is a new field with new systems, the rate of change among the offerings is at least monthly, but often even weekly and daily. The following are some of the current Docker orchestration systems:
Kubernetes by Google
Kubernetes manages a cluster of nodes that run containerized applications. This means that Kubernetes can take multiple computers and treat their combined capacity as one large computer on which Docker containers can run. Kubernetes was developed by Google and is used by Google to manage its own infrastructure. Currently, Kubernetes is supported on Google Compute Engine, Rackspace, Microsoft Azure, and vSphere environments.
Elastic Container Service
Like Kubernetes, Elastic Container Service (ECS) manages a cluster of compute instances as a single resource to locate Docker containers. ECS is a free service integrated with the Amazon Web Services (AWS) API infrastructure.
Rancher (http://rancher.com/)
Rancher creates and manages a purpose-built environment for running Docker. Rancher takes computing resources and loads its own software system (RancherOS) to run Docker. Rancher creates a networked cluster environment that provides load balancing, service discovery, and cross-host networking.
Docker
Docker itself provides the following orchestration tools:
● Machine. Docker Machine manages the configuration of Docker software on individual computers, or groups of computers (called swarms in Docker).
● Swarm. Docker Swarm pulls individual computer resources into a single integrated pool of resources that can run Docker containers.
● Compose. Docker Compose allows users to take small multiple Docker containers that would support individual services, and deploy those services in a microservice architecture with a single file.