Training Categories
Click on a link to see more about each course.
Course Outline
Dates Offered (click on the desired date to register):
This course is currently not scheduled for the near future.
Linux Clusters
Course Number: LIN3
Course Description: Today, building a Beowulf, a massively parallel computer that performs similarly to a supercomputer, is being done in universities, in corporations, and by Linux aficionados who just want the experience of building a cluster. This is being accomplished at a fraction of the price with all the benefits of speed and performance. This course will introduce participants to the intricacies of the implementation and usage of Linux clustering using Beowulf. This course covers the installation and setup of the Beowulf software in any Linux distribution, Beowulf network drivers, management and administrations of a Linux clusters and the tools used in this environment.
Objectives: Participants will learn how to: Use Beowulf multi computer architecture, Setup and Install a Beowulf to create a Linux clustered system, Manage and administrate a Linux clustered system, Install, tune the Beowulf network drivers · Build applications for a Linux clustered system, Select what components are needed to setup the Linux clusters, Setup a file system for a Beowulf Linux cluster, Determine the Limitations of a Beowulf Cluster, Build a Beowulf cluster, Create programs in MPI
Benefits:
Target Audience:
Prerequisites: Participants should have recent experience installing and administering Linux, practical experience in Linux system administration, Linux network management and knowledge of Linux (or UNIX) system commands. Programming experience in C is required.
Course Length: 3 – 4 days
CourseTopics:
Introduction
Basic Concepts,
Background,
History
Nasa HPCC program,
PcPC (Pile of PC’s),
Why Clusters
Cluster Concepts,
Networking Concepts,
TCP/IP Addressing,
Parallel Programming Systems,
Designing Clusters,
Beowulf Architecture characteristics,
Design considerations,
Hardware for Clusters,
System Performance Analysis,
Small scale software provides big scale,
Disk Selection
Network Selection
Network Structure
Network Software and setup
Maximum bandwidth
Cluster Configuration
Parallel filesystem
Scalability
Data Access for Clusters,
Messaging Systems,
Queuing Systems
Compilers
Other Tools
Building Clusters,
Selecting the Place,
Preparing the Environment
Building the Nodes,
Beowulf networking drivers
Assembling Custom Systems
Installation and Cabling,
Software Installation and Configuration
Installation Overview, Linux Support,
Unified /proc Filesystem
Managing Clusters
Basic Tools
The Cluster Management System,
Structure of the Management System, The Cluster Management Page,
User Administration,
Group Administration
Project Administration,
Batch Queues
Accounting,
Cluster Maintenance,
Job and resource management,
Scheduling parallel jobs on clusters,
Parallel program scheduling techniques,
Other Management Interfaces,
Other Useful Tools,
Performance Impact,
Tools and Libraries for Parallel Programming,
Development Tools,
Parallel Development Environments,
Parallel Libraries,
Debugging/Profiling Tools,
Using MPICH to run parallel programs, capacity and to throughput test programs,
Online Documentation,
System Extension,
Monitoring with PCP,
Programming in a Parallel Environment,
Programming Models,
Programming a Clustered System,
Language Selection,
Where to Go from Here,
Summary