completion of one of the world's most powerful supercomputing clusters --
and Canada's fastest system. With the ability to compute more than 600
million calculations per second, the Linux* system is part of an advanced
technology infrastructure supporting research on proteins' roles in causing
diseases such as cancer, AIDS and depression.
Located at MDS Proteomics' new global research headquarters in Toronto and
research facility in Denmark, the newly completed system consists of three
clusters of IBM eServer** xSeries** 330 systems running Linux, with 100
servers in each cluster. The xSeries 330 Intel-based server -- about the
size of a pizza box -- is the most powerful thin server in the industry.
It employs self-healing and self-protective technologies and offers high
availability, reliability and ease of use, with maximum server density.
The Linux supercluster is part of an integrated computing platform deployed
at MDS Proteomics to identify, analyze and compare proteins and protein
interactions. Other systems -- which support both wet-lab (real life) and
in silico (computer simulation) analyses -- include:
-- A network of ultra-sensitive mass spectrometers, which can identify and
analyze thousands of proteins each day.
-- A network of IBM UNIX-based servers, including an IBM SP supercomputer;
eight IBM pSeries** H80 servers; three IBM pSeries M80 servers; and two IBM
p680 servers. These servers support:
MDS Proteomics' new Laboratory Information Management Systems (LIMS) --
the backbone of the company's day-to-day research operations;
MDS Proteomics' in-house databases, including BIND*** (Biomolecular
Interaction Network Database), a repository that stores descriptions of how
biomolecules interact and how they participate in biological pathways to
carry out specific cellular functions; and
Software applications for identifying proteins, bioinformatics
analyses, and drug discovery, including BLAST, which is designed to
compare existing protein and nucleic acid sequences against National Center
for Biotechnology Information (NCBI) databases of sequences; SeqHound, a
protein sequence finder; and MDS Proteomic's Small Molecule Interaction
Database (SMID)***, which contains information on small molecule
-- A 120-port Foundary Gigabit Ethernet -- one of Canada's largest fibre
optic network switches -- to enable the servers to work on analyses in
-- IBM DB2 Universal Database, which provides a scalable relational data
management system for managing complex, heterogeneous proteomic data.
IBM Enterprise Storage Server (code named "Shark), a high-speed solution
for storing, accessing and retrieving the enormous quantities of protein
sequence data generated daily by the mass spectrometers.
-- IBM Linear Tape Open tape library system, which provides 28 terabytes of
backup storage capacity.
-- Fault-tolerant, cluster-based software to manage system and network
availability and reliability, including Tivoli Storage Manager** (TSM).
MDS Proteomics also plans to use IBM's DiscoveryLink** software, which
provides data integration capabilities to seamlessly integrate proteomics
data from a variety of sources, formats and file types. DiscoveryLink
enables researchers to consolidate information from many sources into a
"virtual database" to solve complex medical research problems.
* Indicates registered trademark of Linus Torvalds.
** Indicates trademark or registered trademark of IBM Corporation.
*** Indicates trademark or registered trademark of MDS Proteomics.