Home > Workload Solutions > High Performance Computing > White Papers > Dell Validated Design for Government HPC, Artificial Intelligence, and Data Analytics: AI Inferencing Option > Infrastructure servers
Infrastructure servers provide the non-compute services in the cluster including administration and user access. Typically, these servers fall into two roles: management and login nodes. The exact configuration and number of infrastructure servers depend on the cluster size and requirements. For example, a single management node (also called a head node or control plane, depending on context) is sufficient for small HPC clusters. However, if a cluster has High Availability (HA) failover requirements, a second server is necessary. While login servers are not required, separating users from the critical management systems makes administration easier and minimizes unplanned downtime. Some organizations may require locally compiled code. This can be performed on the login server. A typical system has one login server per every 30 to 100 users. Infrastructure nodes can also be used to provide additional services such as NFS.
The baseline configurations for infrastructure nodes are based on the R6615 (AMD) and R660 (Intel) servers. Density is not a concern because infrastructure nodes are a small fraction of the overall cluster, and therefore regular 1U or 2U systems can be used in single or dual socket configurations. Clusters will also typically use matched architecture for infrastructure and compute servers, for both CPU vendor as well as Operating System family. The former can simplify ease of ordering and physical maintenance, while the latter is useful for compatibility of OS images and software package management.