Information Technology Reference
In-Depth Information
Fig. 3.2:
Typical Form of Cluster Grids
A cluster is a homogeneous entity. Its components differ primarily in configuration,
not basic architecture. Cluster Grids are local resources that operate inside the fire-
wall and are controlled by a single administrative entity that has complete control
over each component (Foster and Kesselman 1998). Thus, clusters do actually not
involve sharing of resources and cannot be considered as Grids in the narrow sense.
However, they are usually starting points for building Grids and a first step towards
Grid Computing. Cluster Grids improve compute and storage capacity within a
company (see also Keating 2004).
Example
Life science companies typically have high demand for HPC, e.g. for the compute-
intensive gene analysis. These needs can be met by deploying a cluster. One
example for the application of clusters in life sciences is the biotechnology company
diaDexus (http://www.diadexus.com) that is focused on bringing genomic biomar-
kers to market as diagnostics. diaDexus replaced their large proprietary Symmetric
MultiProcessing (SMP) machines with a Linux cluster (Keating 2004, Entrepreneur
2003). The gen analysis programs were adopted to a cluster environment and then
implemented on the cluster. Compared to the previous system, the new cluster solu-
tion helped diaDexus to “speed the rate of gene analysis for the development of
future diagnostic products from weeks to hours, or by more than 20x” (Entrepreneur
2003).
