Management subsystem overview

The management subsystem in the high-performance computing (HPC) cluster solution uses an InfiniBand fabric that loosely integrates the typical IBM® System p® or IBM Power Systems™ server HPC cluster components with the QLogic components.

Figure 1 shows the use of a host-based subnet manager (HSM), rather than an embedded subnet manager (ESM), running on a switch. Because a host-based Subnet Manager scales better than an embedded subnet manager and has been verified and approved by IBM, the HSM can be used to run a fabric management server.

The following figure illustrates the functions of the management or service subsystem.

Figure 1. Management subsystem

The servers are monitored and serviced in the same fashion as for any IBM Power Systems cluster.

The Cluster Systems Management (CSM) Management Server (CSM/MS) is the central point for managing and monitoring operations for the system administrator. The CSM functions for event management of the switch and subnet manager events, and for remote command processing to the switches and fabric management server can be used from the CSM/MS. However, the system administrator can also perform these functions by directly logging on to the switches or to the Fabric Management Servers or hosts.

The following table is a quick reference for the various management hosts or consoles in the cluster, the intended user (for example, the system administrator or switch service provider), and the networks to which the hosts or consoles are connected.