[Beowulf] hadoop

Eugen Leitl eugen at leitl.org
Tue Nov 27 06:40:29 PST 2012


On Tue, Nov 27, 2012 at 03:01:57PM +0100, Jonathan Aquilina wrote:
> One thing that really caught my eye was seamicros new storage servers up to
> 5 petabytes of storage

Please do not top-post and please trim your replies. Message
below unchanged for illustration.

The Seamicro is not the same thing, since I would be
looking at least for an Exynos 5 with 1-2 GByte
RAM and some GByte flash for each SATA spindle.
You would not just run a distributed filesystem
or even a database, but to process code locally
that uses 1-3 TByte worth of data on said spindle,
with some SSD/RAM caching. If you're moving a
sliding window across a 3d voxelset where you
do some processing, then you would likely to
be able to stream at 100-200 MByte/s, and
process your TByte/node data within nongeological
time spans. There would be very little traffic on
the GBit port, as you would only query adjacent
nodes when you need to get a slice at the surface
of the cube.

In principle this would map very well to a 3d torus,
but GBit Ethernet for many 10 k nodes is reasonably
cheap.
 
> On Tue, Nov 27, 2012 at 2:59 PM, Eugen Leitl <eugen at leitl.org> wrote:
> 
> > On Tue, Nov 27, 2012 at 09:10:32AM +0100, Jonathan Aquilina wrote:
> > > Hey guys I was looking at the hadoop page and it got me wondering. is it
> > > possible to cluster together storage servers? If so how efficient would a
> > > cluster of them be?
> >
> > An interesting problem would be to use reasonably powerful but
> > cheap ARM SoCs in few GBytes onboard RAM and some flash
> > for hybrid filesystems for each hard drive, and cluster
> > them via GBit Ethernet on a very large scale.
> >
> > That would be a custom Beowulf for more storage-related
> > tasks. E.g. an application I have in mind are volumetric
> > datasets with e.g. 8 nm - voxels for biological systems,
> > which are way too large to process in memory.
> > _______________________________________________
> > Beowulf mailing list, Beowulf at beowulf.org sponsored by Penguin Computing
> > To change your subscription (digest mode or unsubscribe) visit
> > http://www.beowulf.org/mailman/listinfo/beowulf
> >
> 
> 
> 
> -- 
> Jonathan Aquilina
-- 
Eugen* Leitl <a href="http://leitl.org">leitl</a> http://leitl.org
______________________________________________________________
ICBM: 48.07100, 11.36820 http://www.ativel.com http://postbiota.org
8B29F6BE: 099D 78BA 2FD3 B014 B08A  7779 75B0 2443 8B29 F6BE


More information about the Beowulf mailing list