Glossary

The ProLiant DL series servers used in the CP4000 platform employ the AMD Opteron 32/64-bit CPU.

Application program interface.

Basic Input/Output System.

A set of servers joined together using various interconnect technologies to form a supercomputer.

Administrators can use the CLI to automate job, job queue, and node operations. These operations can also be scripted.

Any discrete device that is used to build the cluster, such as a server, network switch or high-speed interconnect.

A server that runs end-user programs. A bundle of compute nodes requires one head node.

Dynamic Host Configuration Protocol.

A device that connects the nodes into the administration networks. These networks facilitate management traffic between nodes and are used to perform functions such as booting.

The customer's local area network (LAN) or wide area network (WAN), to which the cluster can be connected through its internal administrative network.

A method of assembling cluster components into modules and temporarily connecting the modules to form a running, testable cluster. The cluster is partially disassembled into its modules for shipping.

A network topology used by the interconnect. The topology is characterized by a link capacity that is progressively larger closest to the root of the tree than it is at the leaves, making the network non-blocking.

A cluster with more than 128 nodes, requiring interconnects that are organized into a node-level and top-level hierarchy to increase the number of links.

A network topology consisting of a single root device with several individual connections to multiple branch devices. The cluster's internal networks are organized as a flat tree.

An interconnect protocol and connection type supported by the hardware that is based on HP ProCurve and Myricom Myrinet technologies.

A server that provides management functions and user interface for the cluster. One head node exists in the cluster.

PCI interconnect card used in InfiniBand configurations.

An interconnect protocol and connection type supported by interconnect technology from Voltaire that complies with the specification defined by the InfiniBand trade group.

A module containing one or more interconnects.

The Job Scheduler service runs on the head node and is responsible for job submission, queue management, resource allocation, and job execution. This service communicates with the Node Manager service that runs on all compute nodes.

A rack-integrated workstation (keyboard, video, and mouse) enabling you to administer the cluster locally when necessary for tasks such as upgrading firmware and installing software. Most runtime administration tasks are performed by using a remote login from a client.

A connection or device that is part of the interconnect, enabling you to connect to it and administer it. You use the control processor for tasks such as assigning TCP/IP addresses, loading firmware, or running interconnect diagnostics.

Each node has at least two integrated network interface cards (NICs). One NIC might be dedicated to the server's console. These NICs are used for connections to the internal networks.

In federated clusters, an interconnect that has connections to nodes.

The enclosure management hardware, software, and firmware that is used to support all of the managed devices contained within the enclosure.

Preboot Execution Environment

RIS allows PXE BIOS-enabled computers to remotely execute boot environment variables.

In federated clusters, an interconnect that provides links to other interconnects, multiplying the number of possible pathways between nodes to create a full bandwidth fat tree topology.

The organization of a network. HP Cluster Platforms contain networks based on flat tree and fat tree topologies.