deployment phase has become much easier to
implement. Containers can be destroyed and restarted
with the new version of code within seconds. Container
orchestrator solutions, such as Kubernetes, DC/OS, and
Docker Datacenter, can be used to simplify this process
even further. By using the container orchestrator APIs, a
call can be made to roll out the new Docker containers
with the new features—and there is zero downtime.
Several strategies can be used to ensure zero downtime
while deploying containers; changing only one container
at a time from a pool of many instances is one of them.
In most situations, there are several environments
available at the deployment stage of the pipeline, but at
least two are present in all cases: staging and production.
In the staging environment, final preparations and
manual testing by the product team takes places to
ensure that everything functions as expected. Approved
changes in the staging environment automatically get
deployed into the production environment.
The CI/CD pipelines described so far can be adapted for
network management and configuration. Taking
advantage of infrastructure as code, network
configurations can be stored in version control systems,
and build servers can be used to integrate with solutions
such as Cisco CML/VIRL (Virtual Internet Routing Lab)
to dynamically instantiate virtual networks, test the
configuration changes in the virtual network, and, if the
tests pass, use automation solutions such as Ansible to
perform the desired changes in the production network.
A working network management CI/CD pipeline could
include the following:
Github.com as a cloud-based version control system or Gogs as a self-
hosted Git option could be used to store the configurations.