Compute nodes

Compute nodes are physical or virtual machines where the container workloads run. You need to install HyperScale on each of your compute nodes. Compute nodes operate in a compute plane that isolates the compute nodes from the data nodes. This isolation helps you to separate container management tasks from resource-intensive data management tasks such as backup and recovery tasks and snapshot management.

Data nodes

Data nodes are physical servers where storage is managed for the container workloads that run on the compute nodes. Data nodes are used to store point-in-time snapshots of container volumes. You need to install HyperScale on each of your data nodes. Data nodes operate in a data plane that isolates the data nodes from compute nodes. The data nodes are responsible for protecting data, ensuring high resiliency of data, and helping you to meet the required Quality of Service SLAs for storage performance and data availability.

HyperScale cluster

A HyperScale cluster contains compute and data nodes that are configured to communicate with each other over public and private network. To form a cluster, you need to configure at least two compute nodes and one data node. For network connectivity, at least one NIC must be configured on each of the nodes. You can scale the cluster to include a maximum of eight compute nodes and four data nodes with a maximum of two NICs configured on each of the compute and data nodes for network connectivity. The public network provides external connectivity to all the nodes and to the hosted containers. The compute nodes also use the public network to exchange control messages in the control plane. The private network is used to route data traffic and does not have external connectivity. It is isolated from all other networks and is used exclusively for data management tasks such full and incremental backups, data resiliency, and recovery. If only a single NIC is configured, both control and data traffic is routed over the single public network.