A Kubernetes node is a virtual machine capable of running Kubernetes workloads. Node pools can easily be provisioned for your cluster through the Web UI or Terraform.
Nodes should be regarded as ephemeral as the node disk is wiped during cluster upgrades. Therefore you should avoid using hostpath volumes whenever possible. Note that this doesn't impact PVC provisioned volumes, which is the recommended way to provision storage.
All nodes run on modern AMD processors. As the cores are shared the nodes can be scheduled to run more efficiently which allows us to offer much lower prices at comparable performance for the average use case.
Nodes are configured with one NAT interface for internet traffic, and one VXLAN interface for inter-node traffic. Only egress traffic over the NAT interface will count towards your nodes monthly traffic allotment. We currently don't offer any support for fixed outbound addresses or per-node publicly routable IP addresses.
Part of the memory and CPU allocation of each node is required for critical processes and daemons such as init system, kubelet, CSI- and CNI daemons. Under normal operation these services will reserve around ~400MiB of memory, meaning that a 2vCPU/4GiB node will have around 3700MiB of usable memory.
We use a zero trust security model. All nodes communicate with api-servers over mutual TLS with Symbiosis handling the distribution, signing and renewal of node and control plane certificates.
Node images are automatically encrypted on rest using LUKS.