DB2 EEE on a 216 CPU Beowulf

[Christopher Hogue]

Hi Folks,

Just a note to let you know that the IBM Toronto Lab have sucessfully
installed and tested the beta of DB2 Universal Database for Linux
Enterprise Extended Edition (EEE) on our cluster.  

The DB2 folks made it through a standard ad-hoc query set on a small
test database distributed across 108 dual PIII 450 nodes.  Nothing
spectacular to report yet other than that it works!  We will be trying
it out on a larger database shortly - we have about 5Gb free on each
machine's hard disk right now - about 500Gb total available to DB2.  

We plan on loading a Medline XML dump (about 60Gb) onto it over the next
month and see what it can do with real data.

Hmmm - Back of the envelope calculation - with 216 45 GB IDE drives we
could install a pair in each node, mirror them with software RAID and
have 5 Tb of redundant storage for around $40K.  Apparently DB2 can also
put data across the nodes in a redundant fashion, so perhaps the RAID
isn't necessary, and it costs half that.  Hmmm...   

I will keep you posted on the results as we get this a bit more tested
and as we learn more about what a beowulf cluster can do as a big
database machine.

DB2 EEE beta:
http://www.ibm.com/software/data/db2/linux/eeebeta/

Our cluster:
http://bioinfo.mshri.on.ca/yac/

---------------------------------------
Christopher W.V. Hogue, Ph.D.
Samuel Lunenfeld Research Institute
Mt. Sinai Hospital 600 University Ave.
Toronto Ontario Canada M5G 1X5
(416) 586-4800 xt2866
fax (416) 586-8857

hogue at mshri.on.ca