Slashdot Mirror

"but if your Boss wants a cluster setup by tomorrow, and you only have a Windows Admin on hand"

Option 1) Windows admin calls Microsoft, waits on-hold for three hours to hear that he needs to become a registered enterprise customer first, and that then he can request a quote, after which .... yada yada the 24 hours are gone. No alternatives.

Option 2) Use one of the so many turnkey solutions available for Linux.

Check out www.beowulf-underground.org That is the place for everything beowulf. It is run by the guys in the Parallel Architecture Research Lab at Clemson University.

'Supercomputer on a CD' software is supposed to make it so easy to put a Beowulf cluster together

I've wanted to build a cluster for a long time. I was given that chance at school & work. School(U of C) allowed me to do it as a term project that is still ongoing. Work allowed me to use a pile of spare machines that were waiting for new users.

So, early one Saturday morning, I sit down with the O'Reilley book "Building Linux Clusters" and the CD that came with it. I followed the instructions in the book, and was frustrated beyond belief. The CD contained all the Beowulf software that was required. The downside was that the software had all been thrown over top of Red Hat 6.2. Being the second time I had ever installed Red Hat, I wasn't sure what the magic sequence was to get everything to work.

I first had problems with unsupported video cards... I tried 3 different cards. Each time, I needed to re-install Red Hat. Why wouldn't it let me install all the drivers for all the cards?

DHCP? Why? You only need to set the IPs once. Don't force me to do this... oh wait, I don't know Red Hat's weird config script structure... sigh.

The book mentioned nothing about re-compiling the kernel. But, in order to add support for the network cards, this is what I needed to do. Oh wait, where are the kernel sources?

This is when I got sick of this "wonderful" Beowulf CD.

I went to Slackware 7.1.

Installed it on the master no problem. Enabled frame buffer support for the video card so that it would work on ANY video card. Enabled native support for the network cards I was using.

Next step, I went to THE beowulf site, did a search for PVM and PVMPOV. I downloaded all the source code I needed.

Now, without the help of the book, I was at a bit of a loss. Luckily, there was this site that explained EVERY STEP in about a page and a half. The how-to was written by Christopher Johnson and I must say, he did an excellent job. I found only one thing that was lacking, you may also need to set PVM_DPATH=/home/pvm3/lib/pvmd in your profile.

Now the purpose behind all of this was to get PVMPOV running, well, with a little searching, I was able to find everything I needed here.

Conclusions:
Use a Linux distro you are used to.
Get a book if you want to know the theory.
Always remember that some PVM Books are free.

I hope this will help someone out there.
Beware TPB

I think the last version of Extreme Linux was (searches for his Extreme Linux CD) is based on RedHat Linux 5.0 - it's a little out of date now - code has moved on considerably.

For you I would like to recommend some reading:

Building Linux Clusters by David HM Spector published by O'Reilly, (hmmm site seems to be down, come back later, or check Google cached version)

This book comes with a CD together with clustering software. It also comes with step-by-step instructions. I believe, however, that there are some errata, which means that some hacking will need to be done to get your cluster online.
It also goes through some aspects of choosing hardware etc...

A more in-depth resource, without step-by-step instructions, but with in-depth discussions on granularity of Beowulf systems and whether they are actually good for the tasks you have in hand is:
How to Build a Beowulf, A guide to the implementation and application of PC Clusters by the MIT Press

Also check the The Beowulf Project Site and the The Beowulf Underground Site

Have fun!
---

Then you're doing something wrong... a 10-node PIII 550 Mhz System can stomp all over any RS/6000 out there. I administer a 16-node PIII-500 Beowulf and we routinely get Flops competitive to an Origin 2000 (for 1/30th of the price).

It all depends on: a) your application, b) your choice of networking hardware and software. Flops-intensive software will fare *very* well on a Beowulf. I/O-intensive stuff (like retrieving huge amounts of data off an RDBS) would most likely be better off on a mainframe.

Check out this site and tune up your cluster...

engineers never lie; we just approximate the truth.

Further it looks as if it is an 8 node cluster. That isn't a very large cluster. It would be nice to think that this would scale similarly to other super computers, but I can't see the performance of any network approaching the speed of something like a crossbar switch anytime soon.

Perhaps I'm missing something, but at $100k I think it would be safer to build your cluster, or purchase a tried machine in a similar price range. (such as the SGI origin)

It is nice they have some packaged software but is it better than what we can get at http://www.beowulf-underground.org/?
Perhaps, perhaps not. I hope to see what people who have used this machine, other supercomputers AND beowulfs have to say about, perhaps than we can decide if this is a good value.

Doesn't http://www.paralline.com/ sell similar machines anyway?

For now I just don't see the value in it.
But thats just my opinion, I could be wrong. :)

We build turn-key Beowulf systems for a living. You may want to consult:

Beowulf.org

Beowulf FAQ

Beowulf Underground

Beowulf Quick Start

In addition, you may want to contact us directly about helping you convince your management that turn-key and supportable Linux Beowulf Clusters are available and do useful things. Take a look at our success stories.

Doug Eadline,
deadline@plogic.com
Paralogic, Inc.

Look at the Beowulf Underground for an excellent compilation of links, resources and software.

Also, if you're not gonna need some of the more specific software (kernel patches, ethernet channel bonding and the like that usually come in RPMs) but just want to implement generic MPI or PVM, I'd go with Debian next time instead of RH6.0, purely for maintenance reasons.

The big question you have to ask yourself though, is what kind of application you want implemented, and build the cluster to match it... if you don't have an application in mind, then you probably don't need a Beowulf...

Can anyone point me in the right direction...

The Beowulf Underground site is a good starting place.

D.
..is for Deranged.

You don't need the rpms from beowulf.org unless you want to use extensions to the standard kernel such as channel bonding or distributed pids. Otherwise, just install your favorite distribution on several machines, grab MPICH or PVM, and start experimenting.

Feel free to send questions my way if you need help with this. Or check out any of the documentation at http://www.beowulf-underground.org/.

-jason

Here are some cool links:

Beowulf Underground


Extreme Machines