IBM Clustering Solutions

IBM is a pioneer in the clustered computing field, and continues to dominate the Top 500 supercomputer list. In addition to creating cluster solutions for HPC (high performance computing) environments, they also offer solutions appropriate for BPC (business performance computing) workloads.

IBM offers fully-integrated clustering products that combine IBM’s servers, storage and networking products with powerful cluster management software and best-of-breed third-party technology. By offering an integrated solution, customers are provided with a solution that is fully interoperable, centrally managed, and all under one warranty.

IBM Cluster System 1350 - this Linux® clustering solution allows users to choose Intel® Xeon®, AMD Opteron™, PowerPC®, or POWER5+™ processors, and control it with a single point of management with the IBM CSM(Cluster Systems Management) tool.

IBM Cluster System 1600 - leverage POWER5™ & POWER5+ processing power to create a highly scalable system that can run AIX5L™ or Linux. Interconnect nodes on existing Ethernet or Gigabit Ethernet technology, or use IBM’s High-Performance Switch, Infiniband, or Myrinet.

Clustering Software

CSM

CSM is a powerful clustered management tool that provides administrators with the ability to monitor hardware and software events from a single point. With CSM you can group nodes and apply commands and rules to individual nodes, groups of nodes, or entire clusters. Monitor events remotely, and automate recovery with this node control software that provides rapid problem detection and resolution capabilities. Other features include:

Distributed management - a persistent repository of information about each node, and current status.

Event response resource manager - run a rich set of predefined command or scripts in response to users events.

Remote hardware control - remotely reset or power nodes on or off.

Configuration file management - a repository of files that are common between nodes can be synchronized across the cluster as changes occur.

Distributed shell - run commands or scripts remotely on all nodes in the cluster, and combine output from multiple nodes.

GPFS

The General Parallel File System (GPFS) is a high-performance cluster file system that breaks files into blocks across multiple cluster nodes, and can read/write these blocks in parallel. By providing multiple data paths, you can eliminate single-point-of-failure issues and maximize availability. Based on a shared disk model, GPFS reduces overhead access to disk resources not directly attached to the application nodes, and provide data coherence across nodes.

ESSL & Parallel ESSL

Designed for math-intensive scientific or engineering applications, this collection of mathematical subroutines is specifically tuned for IBM Cluster Solutions, and support C, C++ and Fortran applications. Parallel ESSL provides parallel computing capabilities with the Message Parsing Interface.

Parallel Environment

This powerful AIX5L tool provides the ability to quickly develop, deploy and tune parallel applications in C++, C or Fortran. Utilize clustering technologies like distributed memory and message passing to various nodes in this development and execution environment.

LoadLeveler®

Manage jobs across clusters with this dynamic workload scheduling tool designed to maximize resource utilization while minimize compute times.

You can develop dynamic policies that schedule jobs by job priority, job requirements, resource availability, and any user-defined rules. This tool also provides consolidated accounting and reporting.

North American Systems provides a variety of services to complement your cluster solution, helping you determine ROI and set project scope, configuring and staging an appropriate solution, and helping you install, deploy and tune a highly effective & integrated cluster computing solution that meets your business needs.

---

•  Print This Page