Geoscience Reference
In-Depth Information
main potential for insight on their own, merely not from context. Capability computing will
usually not be the first association with information systems.
2.2 Capacity computing
The more interesting along with the development of the next generation of system
architectures is the opening of capacity computing for complex information systems. These
systems provide capacity for many small or medium sized jobs. Job processing of several
parallel jobs is suitable for these resources. The single job itself may be used for parameter
studies, design alternatives, exploring pre-development stages, and in general these jobs on
their own have less potential for insight. Making use of capacity computing resources many
instances of compute jobs will run on a resource. This is what we need to enliven conventional
information system implementations with new features, leading to new insights.
3. Computing resources
3.1 Paradigms
There is a number of different paradigms and resources that can be considered for this
purpose and used from various geosciences disciplines. The topmost category for High End
Computing (HEC) are High Performance Computing (HPC) and Supercomputing resources.
These will in nearly all use cases be used in a non distributed manner. The lower end,
Distributed Computing (DC) and services computing resources, are for example built on base
of paradigms like Sky Computing, Cloud Computing, Grid Computing, Cluster Computing,
Mobile Computing.
3.2 Performance pyramid
What do we have to expect to be the machines behind these paradigms? The answer is
a performance pyramid. The performance pyramid for computing resources shows the
following structure.
Top sector:
• International supercomputers,
• National supercomputers,
• Regional supercomputers.
Medium and bottom sector:
• Local and dedicated compute servers and clusters,
• Workstations,
• Mobile devices.
4. Computing obstacles
The obstacles for operating complex resources with interactive systems with efficient and
effective operation can be overcome while reacting on several levels, architectures, hardware,
frameworks and middleware, applications, energy consumption, competence resources,
consultancy, and support.
Search WWH ::
Custom Search