Re: question about cluster

Mulyadi Santosa <[email protected]> Sat, 28 Jul 2012 11:20:24 +0700
Newsgroups gmane.linux.cluster.ssic.user
Message-ID <CAGdaadbgCVwXqN8hig1EjDPvvzb8VG0XgyM+eA-B32DO5C2nxA@mail.gmail.com>
Hi :)

Please see my reply inline below...

On Sat, Jul 28, 2012 at 2:42 AM, Mr U <[email protected]> wrote:
> thank you for reply

You welcome :)

>
> after sending that mail i understand openssi give me am computer with many cpu not with big cpu,

Yup, "many".... think of it like PC with 96 CPUs, but split across
physical computers.

> my main requirement is my cluster finish analyzes as soon as possible.

there are many ways to achieve it, IMHO. One way I can think very
quickly now, is by using software called batch scheduler, let's say
Sun Grid Engine.

Queue the jobs that works on each of your MRI files, let's say 16
jobs. Configure the batch scheduler to spread the job on your nodes.
That way, in my opinion, you will achieve parallelism. Again, with the
help of cluster file system, you will easily manage those MRI files
from single place.

-- 
regards,

Mulyadi Santosa
Freelance Linux trainer and consultant

blog: the-hydra.blogspot.com
training: mulyaditraining.blogspot.com

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/