Date: Fri, 3 Feb 2006 12:48:53 -0600 From: Joshua Kampmeier <unixpunx83@gmail.com> To: freebsd-questions@freebsd.org Subject: Looking to build an HPC cluster. Message-ID: <591590c30602031048g1a9c483crc1aa76b651fead33@mail.gmail.com>
next in thread | raw e-mail | index | archive | help
Hey Everyone, I have recently become increasingly interested in building a cluster at home(for the knowledge) so that I may be able to incorporprate one at work. I have 4 identical machines (i know this isn't required, but it will make things easier on my part). They are just Gateway desktops - 933mhz 128MB RA= M 20GB HDD. One of them has 256MB RAM and an additional 60GB HDD that I was planning to use a an NFS export for /home. I have been looking for documentation/books for FreeBSD clustering, but there doesn't seem to be a lot on the subject. I have seen the Fellowship and read through that. If anyone can point me to any books, or documentation online, that would be great. I am looking at using Open-MPI for the clustering. I hope that's a good choice? Another question I had is: In the techTV show from...a while back... with Brooks and and Matt Olander doing the FreeBSD cluster, Matt had mentioned that 'The Matrix' was rednered on a FreeBSD cluster. In regards to that, ca= n a FreeBSD cluster provide processing power to Windows-based applications, o= r were they using some kind of unix-based graphics rendering tool? To get a little more in depth: Our enterprise software at work is 'H.M.S. for Windows' from HAB, Inc. It is for public housing operations management. The thing is, this software is all done with Gupta SQLbase and Centura... Our current server is a dual-xeon with 3 GB of RAM, serving about 100 clients. Most of the time, it is sufficient. Sometimes, during the early afternoon i= t can become rather sluggish, and I was just curious if there would possibly be any way to push some of the processing poer for generating reports and whatnot to a small cluster running FreeBSD. I have feeling this ins't possible... but I know nothing about clustering, so hopefully someone can tell me that I have a good idea, or that I'm barking up the wrong tree entirely. Thanks
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?591590c30602031048g1a9c483crc1aa76b651fead33>