xCAT

Overview

Requirements

Download

FAQs

Forum

Reviews

Update: April 28, 2006

What is xCAT?

xCAT (Extreme Cluster Administration Toolkit) is a toolkit that can be used for the deployment and administration of Linux clusters. Its features are based on user requirements, and many of its features take advantage of IBM xSeries hardware.

xCAT makes simple clusters easy, and complex clusters possible, by making it easy to

• install a Linux cluster with utilities for installing many machines in parallel
• manage a Linux cluster with tools for management and parallel operation
• set up a high-performance computing software stack, including software for batch job submission, parallel libraries, and other software that is useful on a cluster
• create and manage diskless clusters.

xCAT works well with the following cluster types:

• HPC: High Performance Computing Physics, Seismic, Computational Fluid Dynamics, Finite Element Analysis, Weather, and other simulations; and Bio-informatic work
• HS: Horizontal Scaling Web farms, etc.
• Administrative: Not a traditional cluster, but a very convenient platform for installing and administering a number of Linux machines
• Windows or other operating systems: With xCAT's cloning and imaging support, it can be used to rapidly deploy and conveniently manage clusters with compute nodes that run Windows or other operating systems.
• Other: xCAT's modular tool kit approach makes it easily adjustable for building any type of cluster.

The Change Log (PDF) contains details of the updated version.

Get Adobe® Reader®

How does it work?

Some of xCAT's features are as follows:

• Support for Redhat and SuSE distributions on all node types, as well as some other popular Linux distributions
• Support for RedHat and SuSE on ia64 on all node types
• Support for any operating system on compute nodes via support for operating system agnostic imaging
• Remote Power control (on/off/state) via IBM Management Processor Network and/or APC Master Switch
• Remote software reset (Ctrl+Alt+Del)
• Remote Network BIOS/firmware update and configuration on much IBM hardware
• Remote operating system console via pluggable support for a number of different terminal servers and "Serial Over Lan" on supported node types
• Remote POST/BIOS console via IBM Management Processor Network and via terminal servers and "Serial Over Lan" on supported node types
• Ability to remotely change boot type (network or local disk) with syslinux
• Automated Parallel installation for RedHat and SuSE
• Automated Parallel installation via imaging with Windows, or other operating systems
• Automated Network installation with supported PXE NICs or via nbgrub on supported NICs without PXE
• Hardware alerts and email notification with IBM's Management Processor Network and SNMP alerts
• Remote vitals (fan speed, temperature, etc.) with IBM's Management Processor Network
• Remote hardware event logs with IBM's Management Processor Network
• Parallel remote shell, ping, rsync, and copy
• Remote hardware inventory with IBM's Management Processor Network
• Torque, PBS, and Maui schedulers, as well as MOAB: Build scripts, documentation, automated set-up, extra related utilities, and deep integration
• Myrinet: automated set-up and installation
• Message Passing Interface: Build scripts, documentation, automated set-up for MPICH, MPICH-GM, and LAM
• Command line utilities for all cluster management functions
• Single operations can be applied in parallel to multiple nodes with a very flexible and customizable group/range functionality
• Support for various user-defined node types.