Cisco Modeling Labs
provides the ability to integrate Docker images into Cisco Modeling Labs
Users are able to
select docker images from public repositories (such as hub.docker.com) or
private repositories. Once downloaded to your Cisco Modeling Labs server, you
are able to design a network topology that will include your docker image.
In Cisco Modeling
Labs, docker functionality is placed inside another virtual machine, CoreOS;
which acts as a host for running docker instances. This is done for two
reasons, for security and to constrain and restrict how many instances you have
running by putting in place memory controls around the resources utilizations
of the various docker instances.
You must install
the CoreOS virtual machine image. This is available for installation from the
Cisco Modeling Labs FileExchange. Please contact email@example.com if you
You can have many
docker instances but you need to be careful with the amount of memory that
docker instances require. Understand that CoreOS is running docker services as
well as the docker instances themselves. There is a limit of 22 docker
instances running at any one time. This limit is set by the number of
interfaces that the KVM supports.
information (interface and routing details) are provided by AutoNetkit using
the build initial configurations function. As part of the simulation launch,
the CoreOS virtual machine is spun up and the docker instance started within
it. The docker instance will appear as if it were directly connected to the
other nodes within your simulation. The neighboring devices are unaware of the
presence of the CoreOS VM that is hosting the docker instances. Each link that
is created in the topology design results in an external tap interface being
created on the CoreOS instance. The CoreOS VM is configured to run with 2Gb RAM
and 2vCPUs. If the amount of memory is insufficient, it can be adjusted using
Resources/Flavors function in the
Workspace Management interface.
There are thousands
of docker images available on public repositories. However, not all images will
run on Cisco Modeling Labs (or any other docker deployment), so care must be
taken when selecting the image.