Kits & Components
Kits & Components
Kits contain multiple components, know how to install and configure themselves, and they ensure software dependencies and pre-requisites are met in an automated fashion. State-of-the-art software Kits packaged with Platform Cluster Manager includes everything you need to have your cluster up and running in no time with the latest HPC software.
An operating system (OS) Kit and the Base Kit are all that is required for Platform Cluster Manager to function, but most cluster users will also deploy the additional Kits referenced below. Kits may be added or removed from a running cluster at any time, and software components can be selectively deployed to different types of hosts with a few simple steps performed through the web management interface (Platform Management Console).
Platform Cluster Manager Kits are often packed with functionality with individual Kits containing several discrete but complementary software tools that tend to be deployed together. By installing and managing software as Kits rather than as traditional software packages, cluster installation and management is simplified dramatically.
Building a Kit is simple. Platform provides clear documentation and easy to follow examples. More Kits are becoming available all the time from the open community (http://hpccommunity.org), ISV partners and hardware manufacturers, and Platform Computing. Please note that the specific list of Kits included in your Platform Cluster Manager distribution may vary depending on where you obtain your software.
Base Kit | The Base Kit includes the core set of advanced Project Kusu facilities and tools that make Platform Cluster Manager clusters so easy to deploy and manage. While cluster administrators may choose to interact with the Platform Management Console (PMC), they can also interact with these base facilities directly. When the Base Kit is deployed a PostgreSQL database is also deployed automatically to manage the state of node configurations and the cluster as a whole. Tools in the Platform Cluster Manager Base Kit include:
|
HPC Kit | The Platform HPC Kit contains libraries, tools, utilities and benchmarks for high-performance computing clusters. Among the tools included in the HPC Kit are:
|
Console Kit | Unique to Platform Cluster Manager, the Console Kit provides the easy to use Platform Management Console web interface and portal. Through this easy to use interface, cluster administrators can manage their Platform Cluster Manager clusters. As additional Platform Cluster Manager Kits are deployed, they become accessible through this GUI automatically providing a single management interface to all cluster operations used by cluster users and administrators alike. |
Ganglia Kit | Ganglia is a scalable statistics collector which monitors node availability, displays system load, network usage, and other resource information for HPC clusters. Ganglia displays detailed information about the entire Platform Cluster Manager cluster and its day-to-day functions, including up-time, available memory, number of processors, CPU usage, etc. Administrators can drill down within the data to view information about each individual node in the cluster. Both server and agent components are included so that Platform Cluster Manager administrators can easily deploy components as needed to a selected set of hosts. |
Platform Lava Kit | Platform Lava is an open-source distributed batch system for submitting jobs and managing the workload on a Platform Cluster Manager cluster. Fully compatible with Platform LSF, Platform Lava lets you manage day-to-day workloads and provides simplified job execution, management, and accounting. When installed, an easy to use job-submission portal and management interface becomes automatically available in the Platform Management Console allowing both users and administrators to submit, monitor and manage workloads with no complicated configuration steps. Both master-host components and compute-host components are included and are configured automatically so that they are ready for use. Platform Lava is an excellent choice for new cluster users who need an integrated, easy to use workload manager that is easily upgradable. |
Nagios® Kit | Nagios is an NMS (network management system) that runs on the Platform Cluster Manager installer node, and monitors your cluster hosts, services, and networks. It provides a web interface that integrates with Platform Management Console to display the collected node information, and to alert you if problems occur within configured thresholds. Nagios monitors a wide range of services on both the primary install and compute nodes in the Platform Cluster Manager cluster. Services monitored include:
|
OFED Kit | The OFED (Open Fabrics Enterprise Distribution) Kit automatically configures OFED support in the cluster. It includes the very latest OFED drivers (v1.4), which support server and storage clustering and grid connectivity, and allows the configuration of IP over IB. Platform Cluster Manager customers using InfiniBand or iWARP hardware will need to install this Kit. |
Cacti Kit | Cacti, included in Platform Cluster Manager, is a complete network graphing solution that uses RRDTool’s data storage and graphing capabilities. Out of the box, Cacti provides a fast poller, advanced graphing templates, multiple data acquisition methods, and various user management features. Cacti integrates with the Platform Management Console and graphically displays almost any kind of data. Cacti is used for built-in reporting in Platform Cluster Manager and also by optional commercial add-on components including Platform RTM, a comprehensive enterprise class cluster monitoring solution. |
Java (JRE) Kit | The Java JRE Kit provides a convenient way to install a standard Java™ 2 Platform Standard Edition Runtime Environment to Platform Cluster Manager managed hosts. |
NTOP Kit | Ntop is an open-source solution to monitor network bandwidth and analyze traffic. It allows you to examine the network patterns of your Platform Cluster Manager cluster and identify “hot spots” and any performance issues. Ntop puts the network interfaces into a passive listening mode so they can watch all traffic coming to and from each interface. The tool plots the data into a database, and then displays this information in the web console, integrated with Platform Management Console. |
Platform LSF Kit1 | With Platform LSF, you can manage and accelerate mission-critical HPC workloads, and intelligently schedule parallel and sequential jobs. Platform LSF provides the capability to solve large, challenging problems while using available computing resources at maximum capacity. The latest Platform Cluster Manager.3 release provides support for Platform LSF 7 update 4. |
Platform MPI Kit | Platform MPI is a language-independent communications protocol used to program parallel compute processes. By using Platform MPI in the development of high performance applications, developers ensure maximum performance for the end user of the applications. Applications developed using Platform MPI will scale more easily as new CPUs are added to the cluster and will port seamlessly when interconnects and other hardware is upgraded. |
DellTM Kit | This Kit, available only in the Platform Cluster Manager – Dell Edition, provides drivers and utilities for the very latest Dell™ PowerEdge servers. It also provides support for IPMI and Dell OpenManage™ (OM). Tools are included to configure the BMC and BIOS settings on cluster nodes during deployment – useful for systems with limited OpenManage support. For systems that have complete OpenManage support, the Dell Kit includes tools to allow configuration of cluster nodes using the omconfig utility provided by OpenManage. This Kit provides extensions specific to Dell hardware such as remote power cycling of servers and console redirection via the Platform Management Console. |
Intel® Kit | This Kit provides additional device drivers and additional support for specific Intel hardware. The Platform Cluster Manager Intel Kit is available in the Platform Cluster Manager – Intel edition only, and supports the very latest Intel Nehalem hardware platforms. |
Intel® Cluster Ready Kit | The Intel cluster checker is a tool to verify the configuration of Linux based clusters and ensure Intel Cluster Ready (ICR) compliance. The Intel Cluster Ready Kit is available only from Intel. |
Red Hat HPC Kit | This Kit, available with Red Hat HPC, is similar to the Platform HPC Kit except that it contains tool and drivers versions qualified and supported with the latest Red Hat HPC software release. |
Red Hat OFED Kit | Like this Red Hat HPC Kit above, the Red Hat OFED Kit is available only with Red Hat HPC, and contains OFED drivers and utilities optimized for Red Hat HPC. |
Mellanox Kit | The Mellanox Kit is only used for clusters deployed with Mellanox hardware. The Platform Cluster Manager Mellanox Kit installs and configures MPI software and the latest OFED drivers, and updates the firmware for the Mellanox InfiniBand HCAs. The Kit supports ConnectX DDR and InfiniHost III DDR InfiniBand HCAs from Mellanox. It provides MVAPICH1 and OpenMPI libraries for supported message passing interfaces and requires that you first install the Platform Base and HPC Kits on the primary install node. |
NVIDIA® CUDA™ Kit | Available as a free optional download for Platform Cluster Manager v1.2.1, a new NVIDIA kit automatically deploys the full CUDA 2.2 environment including drivers, tools, SDKs and debuggers onto Platform Cluster Manager node groups defined via the Platform Cluster Manager GUI as having the Tesla processors installed. This allows customers to easily take advantage of the outstanding floating-point performance offered by NVIDIA GPUs. Read more about accelerating GPU clusters >> |
1 Platform LSF is available as a separate, optional upgrade and is not included in the cost of most Platform Cluster Manager distributions.
2 This product, on its own, does not constitute an Intel® Cluster Ready system, nor does combining it with another product ensure that the complete system is Intel® Cluster Ready, unless the complete system is specifically certified as an Intel® Cluster Ready system.
