Clusters: Mandrake's CLIC and openMosix [LWN.net]

Linux clusters may be one area in which Linux has already achieved World Domination. Clusters running some variant of the Linux operating system are in use at universities, large corporations and research centers worldwide. There are systems made with older hardware, some mix a variety of operating systems and processors in the cluster, but when a special version of the Linux operating system is combined with a set of rack-mounted high-end boxes, Linux clusters are among the fastest and most powerful supercomputers in the world. Clusters present unique challenges for an operating system. Good cluster distributions make it easy to get a large number of boxes up and running, and they make it easy to keep them up-to-date with the latest security and bug fixes. They have kernel patches and other software that enables them to best utilize CPU time. Right now there are several Linux distributors that do this very well, however World Domination requires constant work to maintain. Competition in this arena can only strengthen an already strong product, and keep Linux on top.

Enter MandrakeSoft, who along with partners Bull and INPG/INRIA, has announced the first release of a new Linux Clustering Distribution named "CLIC", a project publicly funded by the French Agency for New Technologies (RNTL). The first CLIC version features rapid deployment, auto-configuration, MPICH, LAM and PVM support, a large number of mathematical libraries, and Netjuggler (a parallelized virtual reality 3D engine). Given the desktop success of Mandrake Linux, we can well imagine that CLIC is easy to use and maintain. CLIC is published under the General Public License (GPL). Users can download the first ISO here.

Also working to keep Linux ahead in clustering is the openMosix Project which has announced the release of openMosix version 2.4.19-6. openMosix is a Linux kernel extension for single-system image clustering, available under the terms of the GPL.

(Log in to post comments)

Clusters: Mandrake's CLIC and openMosix

Posted Oct 31, 2002 14:44 UTC (Thu) by freelsjd (guest, #250) [Link]

I wish I had kept the e-mail response I had received directly from Linus a few years ago. At that time I was advocating the need to get compilers and other software focused toward high-performance floating-point computing. He responded that, no, Linux needed to concentrate on the game market because that is where the most users are.

I think there are several linux gamers out there, but the floating-point world sort of took care of itself.

Linux has still a long way to go

Posted Oct 31, 2002 23:00 UTC (Thu) by jschrod (subscriber, #1646) [Link]

World Domination for Linux in clusters? Where? OK, some universities (cash-stripped as usual) deploy them to give more students and researchers the opportunity to play with them. Great for them, but they are most often not the serious users. Some companies look seriously for Linux clusters since they see the possibility for cost savings for low- to mid-range demands. (E.g., I'm currently involved in benchmarking Linux clusters for crash simulation.)

But, from my hands-on experience with Linux clusters, Sun Cluster (both 2.x and SC3), IBM'S HACMP and POE, Veritas clusters, and (last, but not least) VMS clusters -- believe me: Linux has still quite some way to go until it comes in the region where the major player work today. That holds for all RAS areas (reliability, availability, and serviceability), and for performance.

Linux clusters make headway, and that's a good thing; but they are focused on MPP-style problems which we don't always have and which are hard to program for the application developpers.

In the end, it boils down to simple decisions: If you need your LS-DYNA job (that's crash simulation) or your NASTRAN computation finished the next morning, since it saves hundreds of expensive engineering hours, then you'll go with an IBM Regatta and not with a Linux Cluster. If you have a bit more time, you save the money in hardware and fight with the Linux cluster software.

Clusters: Linux and high performance and high availability.

Posted Nov 1, 2002 13:04 UTC (Fri) by humberto (guest, #70) [Link]

You forgot the mother of all linux clusters:

'Linux will be the main operating system for IBM's upcoming family of "Blue Gene" supercomputers'

'Blue Gene/L, the first member of the family, will contain 65,000 processors and 16 trillion bytes of memory.'

http://news.com.com/2100-1001-963285.html

I recently read that the design specs for Blue Gene call for a processor failure every four days. This will require significant beefing up of linux to handle processor failures.

Clusters: Linux and high performance and high availability.

Posted Nov 4, 2002 16:22 UTC (Mon) by jschrod (subscriber, #1646) [Link]

That's still MPP. What if there is no MPP solution for your problem?

It may not exist or may not be available or may be too expensive to create, that doesn't matter. Quite often SMP is needed in HPC, you can't beat everything with MPP. A good carpenter has more than one hammer in his toolbox...

Joachim