[Bioclusters] Request for discussions-How to build a biocluster Part 4 (batch systems)

[Sylvain Foisy]

Hi,

A reminder: this is coming from a total newbie at this BioCluster stuff. 
it is also to serve as the seed of a tutorial/history-of-building site 
for our creation. I am a total newbie in UNIX administration and 
installation. This is why we will get a system administrator to help us 
out. But I still have to figure out the right questions to ask!!

THE BATCH SYSTEM

We had a look around and decided to go with an open-source system, 
primary because of cost consideration but also for philosophical 
reasons. After looking hard, I found the following solutions:

The Condor project
Parasol
OSCAR
Sun GridEngine

After reading the docs for each, we have pretty much decided either for 
OSCAR or SGE. What should be our criterions for finalizing our choice? 
Any inputs from people with experience with both systems would be 
appreciated.

Also, I would like to know how the user interacts with a cluster. With a 
web page, I figure that a CGI script takes the infos from the user (a la 
NCBI) and turns that into commands for the head to start the jobs. A 
script than takes the outputs from each nodes, sort them according to 
scores and generate an HTML page. Am I wrong? For batch jobs, we are 
thinking in allowing SSH shell access to the head to users. Can a user 
simply send a batch of blastall commands and then get the results or 
should it get access to the BS?

Am I missing something?

This is open for helpful and constructive discussion

Sylvain

++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Sylvain Foisy, Ph. D.
Manager
BIONEQ - Le Reseau quebecois de bioinformatique
Genome-Quebec
Tel.: (514) 343-6111 poste 5188
E-mail: foisys@medcn.umontreal.ca
++++++++++++++++++++++++++++++++++++++++++++++++++++++++