Deploying DTR with high availability requires more than one replica executing all DTR components. We will deploy an odd number of replicas to ensure high availability.
DTR should be deployed on dedicated worker nodes. This will ensure that none of the non-system processes will impact DTR's behavior and vice versa. DTR's processes can take a lot of CPU during scanning and other procedures. Therefore, we will use three dedicated worker nodes. We usually admit DHCP on worker nodes, but we will ask for fixed IP addresses on DTR's worker nodes. We will also require fixed hostnames.
We can deploy the Docker Enterprise platform on-premises or in the cloud. DTR requirements were described in brief in Chapter 11, Universal Control Plane.
To deploy DTR on dedicated workers, these nodes require at least the following:
- 16 GB of RAM
- 2 vCPUs (virtual CPUs)
For production, we will ask for bigger nodes with more resources:
- 32 GB of RAM
- 4 vCPUs
This...