Swansea University’s College of Medicine provides an outstanding infrastructure for computational biology
Built by IBM, Blue C is dedicated to life sciences work, a rarity worldwide for computers of this power.
The HPC facility was initially based on IBM’s pSeries Power4+ technology, configured with a high performance high bandwidth and low latency interconnect with disk and tape storage to exploit parallel programming methods in the Life Science environment. Based on Linpack Test performance, in June 2005 it ranked highly in the top500 list, (1.7 teraflops). In July 2006, the system underwent the first stage of a rolling upgrade, migrating the system from an IBM Power4+ technology based ‘cluster’ to a newer and more powerful offering – IBM Power5+ technology.
The current Blue C High Performance Computing (HPC) cluster Architecture is as follows: The main ‘cluster’ system consists of 22 * IBM pSeries model 575 nodes, each with 16 * 1.9 Power5+ GHz processors and 64 GB of memory, with all nodes using the latest version of the IBM Advanced Instruction Executive (AIX) UNIX operating system – AIX 5L version 5.3 (maintenance level 4). These nodes provide system access (1 node), parallel and serial computation capability (16 nodes), data backup and archive (1 node), storage I/O (2 nodes) and an ‘Interactive Data Services’ system (2 nodes), to host massive medical databases.
The system uses an IBM high performance, low latency switching interconnect (IBM designation ‘Federation’) between all nodes to allow the very high speed transfer of data between processors in physically separate nodes. The system also has an IBM Storage Area Network (SAN) that provides 9.5 TB’s of RAID 5 useable storage, which is accessible by every node within the cluster. Finally, the system is complemented with the use of an IBM Automated Storage Tape Library unit to allow the backup and archive of user data to magnetic tape media, which is managed using IBM’s Tivoli Storage Manager (TSM) product.
Ongoing Blue C projects include numerically intensive analysis of viral genomes, epidemiological modelling of avian influenza and analysis of the genetics of disease susceptibility. Furthermore, projects run by the School of Engineering and Department of Physics on the Swansea University campus have lead to life science projects involving fluid dynamics and imaging. Blue C also supports extremely large clinical databases (Interactive Data Services) for use in on-going e-cohort projects that require large storage
capacity, integrated statistical modelling tools and remote access. Our first major database project, in close collaboration with IBM, was led by Gerry Thomas, and resulted in the development of the complex tissue banking database for the Chernobyl Tissue Bank database project (now located at Imperial College, London, where Professor Thomas took up the Chair of Molecular Pathology in 2007). Our database programme is now centred on the Health Informatics Research Unit (HIRU) and the development of powerful anonymised electronic cohorts of health records.
Click here to view publications enhanced by Blue C
Click here to go back to Supercomputing