The popularity and modular nature of open-source operating systems allows for a great many projects and solutions. As you saw in previous chapters, plenty of information is available about Linux clustering, and there are projects for just about every aspect of clustering. There are parallel clusters, high availability clusters, and soon, you’ll be able to put together two refrigerators running Linux for redundancy!
The following web sites are included as a further guide in selecting the right information for clustering your systems. They are current as of the publication of this book.
You can find other handy tidbits here on how to build large numbers of clusters from scratch, and how to keep them secure:
www.ibiblio.org/mdw/HOWTO/Hardware-HOWTO —. The Linux Hardware Compatibility List.
www.informatik.uni-koeln.de/fai —. Fully Automated Installation (FAI). Kickstart for Debian GNU/Linux.
www.ferzkopp.net/Software/CloneIt/CloneIt.html —. CloneIt is an image-creation system for large numbers of identical PCs.
http://oss.software.ibm.com/developerworks/projects/lui —. Lui is IBM’s open-source cluster installation tool. It reports to be entirely configurable based on client demand of resources.
www.tripwire.org —. The open-sourced version of the security tool. This auditing tool is a must-have for all essential systems.
www.securityfocus.com/bugtraq/archive —. Search for exploits and security issues.
www.bastille-linux.org —. A script to harden Red Hat- and Mandrake-based distributions. This is one of the best scripts to secure your systems, but remember that it’s easy to secure your system beyond what you intended.
www.division17.net (Telecommunications Division 17 Initiative)— . www.csinet.org/. Home of the Construction Specifications Institute. This is helpful for datacenter planning and construction.
www.myricom.com —. Myricom Corporation. Founders of Myrinet, which is a networking technology similar to Gigabit Ethernet.
www.webtechniques.com/archives/1999/08/newton —. Offers technical considerations for planning your computer room.
www.isc.org (Internet Software Consortium)—. Makers of software such as BIND, INN, and an open DHCP implementation.
Linux, being an open-source operating system, can easily handle different file systems and the standard ext2 file system. You want to look into alternative file systems if you’re hosting mission critical applications, just for the journaling, if nothing else:
http://parlweb.parl.clemson.edu/pvfs/ —. The Parallel Virtual File System. This virtual file system is stored on many nodes throughout a cluster and is accessible through each node.
http://oss.sgi.com/projects/xfs/index.html —. XFS, SGI’s open-source journaling file system.
www.coda.cs.cmu.edu/index.html —. Coda distributed file system.
http://oss.software.ibm.com/developerworks/opensource/jfs/?dwzone=open-source (JFS)—. Open-source journaling software for Linux from IBM.
www.jetico.com/index.htm#/linux.html —. BestCrypt for Linux supports encrypted file systems as a regular file system on a normal mount point.
www.gnu.org/software/parted/parted.html —. GNU Parted is a program for modifying Linux partitions. Parted can modify existing partitions, copy data between hard disks, and image them.
Now that you have your cluster up and running, you need to maintain the cluster. If you’re not using a cluster operating system, you might find these programs and scripts helpful to maintaining your cluster’s health:
www.netlib.org/utk/icl/xpvm/xpvm.html —. XPVM is a GUI interface and monitor for PVM.
www.abo.fi/∼mats/CMS. (Cluster Management System)—A front-end interface for a cluster of machines, which executes the Message Passing Interface (MPI). This keeps up-to-date information such as CPU load, processes, and memory utilization across the cluster.
http://smile.cpe.ku.ac.th/research/scms1.2.2 —. The Smile Cluster Management System is another graphical interface that allows for monitoring large clusters and job submission.
www.alphaworks.ibm.com/tech/clusterstarterkit?open&t=gr,p=Clusters4Linux —. The cluster start kit for Linux. Designed by IBM, this allows for the creation and monitoring of up to six nodes from a master node.
www.netlib.org/benchmark/hpl —. A portable implementation of the high-performance Linpack Benchmark for distributed-memory computers.
www.linux-ha.org. (High Availability Linux Project)—Your one-stop shop for all things highly available.
Linux-HA Mailing list archive:
Send a message to [email protected] to subscribe to the list.
Send a message to [email protected] to unsubscribe.
http://oss.missioncriticallinux.com —. Mission Critical Linux.
www.sgi.com/software/failsafe/ —. SGI’s open-source high availability project.
www.linuxjournal.com/article.php?sid=4344 —. High availability cluster checklist.
www.top500.org —. The top 500 supercomputer sites.
www-unix.mcs.anl.gov/mpi/index.html —. MPI Standard.
www.erc.msstate.edu/misc/mpi/mpi-faq.html —. MPI FAQ.
www.lam-mpi.org —. An implementation of the MPI programming environment.
www.cs.sandia.gov/cplant/ —. The computational plant at Sandia National Laboratories. Cplant is a software distribution for Linux parallel clusters.
www.beowulf.org/mailman/listinfo/Beowulf —. Beowulf mailing list. Discussion of topics related to Beowulf clusters.
www.openmp.org —. A multiplatform shared-memory programming interface for C, C++, and Fortran.
http://plogic.com/bcs —. Basic cluster scripts for Beowulf systems.
http://frank.harvard.edu/∼coldwell/diskless/ —. Charles Coldwell’s technique on booting diskless clusters.
www.backhand.org/mod_backhand —. An Apache module that allows for the seamless redirection of HTTP requests from one server to another.
www.supersparrow.org —. As part of the Vanessa project, Super Sparrow allows users to load balance across diverse geographic points by using BGP routing information.
http://gnutella.wego.com/ —. Home of the ever popular Gnutella distributed file-sharing protocol.
www.sun.com/gridware —. Sun’s Grid Engine, an open-source implementation of a distributed clustering environment that is ported to Linux.
http://setiathome.ssl.berkeley.edu/ —. SETI@home.
www.mithral.com/products/cs-sdk/ —. Mithral Software Development Kit is designed to build distributed cluster applications.
www.stanford.edu/group/pandegroup/Cosm/ —. Folding at home, a project dedicated to finding out how proteins self-assemble.
18.191.93.12