While traditional server management interfaces run directly on the BMC and provide a web-based GUI for managing a single server or chassis, openMPMC is deployed as a centralized management platform on a dedicated server or Kubernetes cluster, capable of managing multiple servers and multi-module systems simultaneously. For remote systems in isolated networks, lightweight agents implemented in Go run on local gateway systems and establish outbound gRPC connections to the central openMPMC backend. These agents act as a proxy between the central platform and the BMC Redfish endpoints of the managed systems. For multi-module systems like eHPS, this means connecting to the chassis BMC, which in turn manages the individual compute modules. For EMDC systems, the agents connect to both the backplane BMC and the individual module BMCs. This architecture enables remote diagnostics and control without requiring VPNs or static routes, significantly simplifying network configuration while maintaining security.

Features

A key feature of openMPMC is its bidirectional use of the Redfish standard. The system acts as a Redfish client to interact with hardware Baseboard Management Controllers (BMCs), retrieving detailed hardware inventory, sensor data, and executing remote management actions such as power on/off, reset, setting boot order, or mounting ISO images. Simultaneously, openMPMC serves as a Redfish endpoint itself, allowing third-party systems to query and control resources through a standardized API. This architecture provides a significant advantage: multiple heterogeneous systems can be accessed and managed through this single, unified Redfish endpoint, eliminating the need to connect to each server’s BMC individually (see Figure 2).