[Bioclusters] Request for discussions-How to build a biocluster Part 3 (the OS)

Mike Coleman bioclusters@bioinformatics.org
09 May 2002 00:15:56 -0500


[caveat: I'm studying Linux clustering, but I haven't built one yet, so don't
take this seriously.]


Sylvain Foisy <foisys@mac.com> writes:
> THE OPERATING SYSTEM
> 
> Well, it will be Linux of course. Around here, we use Red Hat 7.2, Mandrake
> and Debian. Any preferences? Any one of them is easier to install/configure
> for clusters?

I'm a big Debian fan, so I'm looking at a way to use it.  One annoyance I see
with various bits of prepackaged clustering systems/software is that *they*
want to choose the distribution for you (usually they choose Red Hat).  This
seems a bit like the tail wagging the dog to me.


>  From what I could figure out from data on the web, it boils down to disk
> vs. diskless. Since part of the database will reside in each and everyone of
> the nodes, am I wrong to assume that the nodes need to have Linux on their
> disk to boot from it? Would it be possible to have the database files on local
> node disk but still have the nodes boot from a disk image located on the
> master node?

Well, the nodes could definitely nfs-boot from a single master whether or not
they actually have local disks.

The question of whether or not completely diskless nodes would work for you
depends on the nature of these databases.  If they are basically read-only and
fit in RAM, you could conceivably grab the data via an NFS mount from a single
master machine.  The idea is that they would get sucked down once after boot
and basically sit in RAM.  If the usage pattern doesn't look like this, you
probably want to have local disks, because a central NFS server will likely
become your bottleneck.

Mike