Welcome to the most active Linux Forum on the web.
Go Back > Forums > Linux Forums > Linux - Networking
User Name
Linux - Networking This forum is for any issue related to networks or networking.
Routing, network cards, OSI, etc. Anything is fair game.


  Search this Thread
Old 07-30-2003, 09:19 PM   #1
Registered: Jul 2003
Distribution: Ubuntu, SUSE
Posts: 79

Rep: Reputation: 15
New to clustering...

Just a few questions about Linux clustering for math-intense operations like data analysis and rendering-

1)Does clustering with, say, 10 older Pentium II boxes create a performance advantage over 1 very fast twin-processor modern machine?

2) Does clustering work with software that was not specifically optimized for such an architecture? For example, could/would a single-user 3D CAD/modelling app speed up if harnessed to a render-farm if it wasn't desgined to do so?

3)How bad is the load on the "controlling" server that coordinates the node operations?

4)In general, is clustering viable for the average Joe with decent Linux networking skills, a low budget, and a couple of power-user apps?

Thanks for the help!
Old 07-31-2003, 10:09 AM   #2
LQ Newbie
Registered: Jul 2003
Location: east coast
Distribution: Slackware and Debian
Posts: 4

Rep: Reputation: 0

This post isnt very technical, just advice. Ok well ive never used clustering for math intensive purposes persay. And I have never worked with 3d CAD modelling apps. BUT i have setup rudimentary linux clusters in the past with RedHat machines. All of these machines were 233 mhz and one was a SMP 300mhz p2, 5-6 machines in total. Clustering was fun and an awesome learning experience but I didnt notice any more 'power' then i had on my 1ghz machines. Of course i wasnt applying the most memory / process intensive applications to my cluster. So my advice would be to research every bit of clustering you can. Theres several types. It may be hard to find this 3d modelling app you want to run that will take advantage of your cluster. Nevertheless you will learn alot from clustering. I would say go with a faster machine for your purposes, a dual 1ghz or something along those lines.
Old 07-31-2003, 11:05 AM   #3
Registered: Feb 2003
Location: The City by the Bay
Posts: 116

Rep: Reputation: 15
The primary advantage of clustering is service availability. Without the traditional PRIMARY-BACKUP architecture where the process of switching from one to the other takes time and can affect multiple users, the 'swarm' will respond immediately to all new requests, and since multiple 'bees' are responding to each request, you preserve data integrity in case of failure.

Cluster is to an application as SMP hw is to a thread. There's less bottleneck.

With a single threaded app so you see an improvement on an SMP rig?

Cheers as a learning experience, for sure. Practically, it may not improve performance


Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is Off
HTML code is Off

Similar Threads
Thread Thread Starter Forum Replies Last Post
clustering mimithebrain Linux - General 2 06-13-2005 09:02 AM
which os for clustering junctionking Linux - Distributions 4 05-25-2005 03:03 PM
Clustering anand_rhce Linux - Hardware 0 12-15-2004 06:01 AM
Clustering alexr186 Linux - Software 3 06-29-2004 06:24 PM
Clustering Stephanie General 5 01-29-2004 12:04 AM

All times are GMT -5. The time now is 09:34 PM.

Main Menu
Write for LQ is looking for people interested in writing Editorials, Articles, Reviews, and more. If you'd like to contribute content, let us know.
Main Menu
RSS1  Latest Threads
RSS1  LQ News
Twitter: @linuxquestions
Facebook: linuxquestions Google+: linuxquestions
Open Source Consulting | Domain Registration