The HP Unified Cluster Portfolio and Cluster Platform Series family of High Performance Computing Clusters

The HP ProLiant HPC Partner Software Suite is a series of software products providing a set of partner developed solutions that complement or expand choices beyond HP's branded offerings. These offerings can be used in stand alone server, custom cluster, or workstation configurations where applicable, in clusters defined as Cluster Platform configurations, or in clustered configuration of mixed server, blade, and in some mixed vendor environments.

These offerings are certified to operate on the HP ProLiant family of high performance designated servers using the latest Red Hat and Novell SuSE offerings, as well as, Windows Compute Cluster Server 2003 in some cases. Please see the Operating System Compatibility Matrices to ensure applicable utilization in this document or on the web at: http://www.hp.com/go/linux and http://www.hp.com/go/microsoft

The HP Unified Cluster Portfolio is based on HP developed and supported hardware delivered as Cluster Platforms depicted above. In addition HP developed and supported software also is provided under the Unified Cluster Portfolio and is referenced in other product offering documentation. Please refer to the following url for HP developed and supported hardware and software - http://www.hp.com/go/clusters

The HP ProLiant HPC Software Suite includes:

• Linux 8 Pack Operating system packages in both 32 and 64 bit offerings for ease of install and administration for both Red Hat and Novell SUSE
• Windows Compute Cluster Server2003 for 64-bit Windows operations - which is referenced in its own product offering but listed here as a component of the HPC Partner Suite.
• Job Managers for 32 and 64-bit Linux or Windows environments
• Cluster Managers for 32 - 64-bit Linux environments

• Updated Red Hat HPC 8 Packs to support RHEL 5 for one and three year subscription services with additional single node licensing for compute nodes. A stand alone Red Hat offering is still required for the Control or Head node
• Consolidated Scali Manage and Scali Manage/Connect and Scali Manage Departmental offerings in cluster node solution offerings for ease of configuration and ordering.
• Consolidated Altair Engineering PBS Pro offerings into a single per Code offering for ease of configuration and ordering

With the introduction of the HP Unified Cluster Portfolio, HP has combined the flexibility of a custom solution with the simplicity, reliability and value of a pre-configured, factory-built product. A wide range of qualified options ensures flexible choices, simple implementation and successful results. The foundation of the Unified Cluster Portfolio is the HP Cluster Platform, standard, integrated implementations of cluster designs that provides customers with a choice of processors, operating systems and interconnects. The HP Cluster Platforms (CP) 3000 and 4000 for ProLiant DL Rack Mount Servers and the CP3000BL and CP4000BL for Xeon and Opteron based HP Blade System. Servers are integrated, customer-ready solutions that enable rapid and confident deployment of High Performance Computing (HPC) clusters.

• The HP ProLiant High Performance Computing (HPC) Software Suite is designed to make more effective use of the HP Cluster Platform series, custom designed clusters, or independent servers attached or used in high performance computing environments. These suites have been certified across the range of HP ProLiant servers and interconnect options used in the HP Cluster Platform series or in custom cluster configurations

The HPC Partner Suite consists of Operating Systems, Cluster Management, and Job Management Kits used in the creation, operation, and expansion of high performance computing environments. The offerings can be ordered for a stand alone server or a cluster of servers consisting of one or more Head Nodes or Control Nodes which act as a central monitoring and deployment server to a number of attached Compute Node servers linked in a common interconnect domain or domains providing cluster interconnect for message passing commands and data, administrative command and control for hardware and software events and performance, and for remote management where local systems administration services are not available or desired. Listed below are the offerings which comprise the HPC Partner Suite of infrastructure applications and their targeted server node placements. These offerings are electronically downloaded from our partner web sites for the most current level of software.

For Partner SKU Operating System Compatibility please visit: http://www.hp.com/go/hpc-resource-room

Factory Integration in EMEA: To order an integrated cluster or for additional requests, send the required specifications in an email to HPTC EMEA cluster team@hp.com (EMEA HPTC Competency Center). A quotation will be provided containing the appropriate configuration and integration charges for approval.

Scali Manage™ enables rapid installation and configuration of clusters including the operating system, middleware, communication modules, third party applications and user data. Companies benefit from having their clusters up and running quickly, reducing the total cost traditionally associated with cumbersome installations. The clusters and applications are brought online quickly and the organization is able to actively predict and prevent system failures through advanced monitoring and alarms. Scali Manage provides a centralized point of installation and ongoing management for clusters, applications, interconnects and industry-standard hardware platforms.

The Scali MPI Connect™ implementation is built on a direct data transfer model that leverages the standardized API intermediate layer based on the Direct Access Transport (DAT) specification. Highly optimized algorithms for message passing through the direct data transport mechanism eliminate overhead and reduce latency. Scali MPI Connect is fully integrated and compatible with the leading underlying inter connect hardware, which enables customers to select the best solution for their applications. The MPI includes a directly attached driver for channel bonded Gigabit Ethernet, as well as 10/100 Fast Ethernet, Myrinet, and InfiniBand enabling high performance clusters to run over commodity interconnect technology as well as specialty cluster interconnect hardware.

A sophisticated, integrated, full-featured Cluster Management software package for clusters using standard Ethernet, Gigabit Ethernet, InfiniBand or Myrinet. Scali supports a specific MPI library which is included with the Scali Manage / Connect package. When using Scali as a Cluster Manager, HP-MPI can also be used as a MPI library instead of Scali Connect. In this case, use the Scali manage offering.

Scali Manage features include:

• Operates in 32 and/or 64 Bit operations in the same cluster
• Operates with both RHEL and/or SUSE OS environments in the same cluster
• Operates across both ProLiant DL and BladeSystem servers in the same cluster
• Operates in a mixed Vendor cluster such as Dell, IBM or most X86 devices
  
• Installation, reinstallation and configuration of the operating system, middleware, communication modules, third party applications and user data
• Ability to manage multiple and mixed interconnect environments
• Resource management, system monitoring and alarms, job management, remote out-of-band operations and software distribution functionality
• Consistent interface regardless of interconnect or cluster
• Monitoring of variables, such as, CPU load, traffic, disk space, swap usage, fan-speed and temperature
• Parallel shell tools for cluster wide command execution
• Interface to any external data source including SNMP
• GUI and command line interfaces for system administrators and application users
• 2D and 3D bar and histogram graphs and charts

Scali MPI Connect features include:

• Dynamic binding to operator- or user-selected interconnects
• Support for key interconnects and standards, including TCP/IP, Direct over Ethernet, SCI, Myrinet and InfiniBand
• Single binary, multiple interconnect architecture enables single instances of ISV applications to be interconnect independent
• Channel bonding and out-of-order package handling in the DET enables near linear scalability
• Support for IA32 and IA64 processing architectures
• Transparency to transient network errors, interconnect resets or change of routing tables
• Dynamic selection of optimal transport medium at runtime
• Failover between interfaces by automatic switching to alternate networks
• Built-in tracing and timing facilities for performance tuning and debugging of parallel codes
• Integration with the Etnus' Total View and GNU gdb debuggers and the Pallas' Vampir MPI analysis tool
• Full compliance with the MPI 1.2 specification from the Message Passing Interface Forum
• Support across both Linux and Microsoft Operating systems 32 and 64 bit implementations

All third-party software products supplied by HP shall be governed exclusively by the manufacturer's license terms and other documentation accompanying such third-party software.

The management systems were inadequate for modern parallel/distributed computers and clusters. From the initial design forward, PBS has included innovative new approaches to resource management and job scheduling, such as the extraction of scheduling policy into a single separable, completely customizable module

Key Features:
User Interfaces - Designed for production environments, xPBS provides a graphical interface for submitting both batch and interactive jobs, querying job, queue, system status, and tracking the progress of jobs. Also available is the PBS command line interface (CLI) providing the same functionality as xPBS.
Job Priority - Users can specify the priority of their jobs, and defaults can be provided at both the queue and system level.
Job-Interdependency - enables the user to define a wide range of interdependencies between batch jobs. Such dependencies include: execution order, synchronization, and execution conditioned on success or failure of a specified other job.
Cross-System Scheduling - provides transparent job scheduling on any system by any authorized user. Jobs can be submitted from any client system or any compute server.
Security and Access Control Lists - Configuration options permit the administrator to allow or deny access on a per-system, per-group, and/or per-user basis.
Job Accounting - maintains detailed logs of system activity for charge back or usage analysis. Custom charges can be tracked per-user, per-group, and/or per-system.
Comprehensive API - a complete Application Programming Interface (API) for sites who desire to integrate PBS with their applications, or have unique job scheduling requirements.
Automatic Load-Leveling - The cluster jobs scheduler provides numerous ways to distribute the workload across a cluster of machines, based on hardware configuration, resource availability and keyboard activity (all of which information is available via PBS).
Enterprise-wide Resource Sharing - PBS does not require that jobs be targeted to a specific computer system. This allows users to submit their job, and have it run on the first available system that meets their resource requirements. This also prevents waiting on a busy system when other computers are idle.
Username Mapping - provides support for mapping user account names on one system to the appropriate name on remote server systems. This allows PBS to fully function in environments where users do not have a consistent username across all the resources they have access to.
Parallel Job Support - supports parallel programming libraries such as MPI, MPL, PVM, and HPF.
Such applications can be scheduled to run within a single multiprocessor system or across multiple clustered systems.
Automatic File Staging - provides users with the ability to specify any files that need to be copied onto the execution host before the job runs, and any that need to be copied off after the job completes. The job will be scheduled to run only after the required files have been successfully transferred. Broad Platform Availability.

NOTE: All third-party software products supplied by HP shall be governed exclusively by the manufacturer's license terms and other documentation accompanying such third-party software.

HP solutions adhere to industry standards, and are deployed at multiple sites with range of open source and community solutions. One of the most popular open source stacks for clustering is ROCKS. If you are a customer that has Linux cluster experience, you can download a RedHat-Rocks combination and configure your own cluster. Rocks consists of a fully integrated and easy to install software bundle designed for high performance cluster Clusters. Everything needed to install, build, maintain, and use a modest sized Linux cluster is included in the suite, making it unnecessary to download or even install any individual software packages on your cluster. HP provides no support for ROCKS. For customers seeking a commercially supported version of ROCKS, Platform Computing offers Open Cluster Software (OCS) utilizing many of the technologies and procedures defined in ROCKS.

HP has done minimal testing for ROCKS to date. If ROCKS is a requirement for a project, please contact the HPC division via the contact form on the web

HP CS provides on-site hardware break-fix support and remote remedial software call center support. CS off-site software services will provide and assist level 1 and 2 support for Red Hat and Novell SUSE Linux and Windows operating system offerings.

• Level 1 is defined as everyday user/system administration issues.
  

Level 2 is defined as issues relating to installation and configuration problems, along with other problems not being solvable by following the Vendor supplied documentation. The CS software support team will work in parallel with the appropriate Vendor and development groups to address level 3 and 4 support escalations:

• Level 3 elevations typically will require patches and modifications to be generated by the Vendor to resolve deficiencies in the product.
  
• Level 4 elevations deal with enhancements in functionality of the product and will typically be included in future releases.
  

The HP ProLiant HPC Partner Solution Suite offerings are covered thru optional partner support agreements which are ordered as additional sku offerings. Customers are asked to contact the respective partner support contacts when the software is delivered either stand alone or pre installed to confirm the service and support linkages and to alert partner support organizations of software receipt. Please contact the appropriate ISV for support conditions.