Queueing Aspects of Integrated Information and Computing Systems in Geosciences and Natural Sciences - Advances in Data, Methods, Models and Their Applications in Geoscience - page 4

As far as High End Computing (HEC) being a genus for High Performance Computing
and various other ambitioned computing paradigms is an issue of national interest for most
countries, reliability and security are the most important factors for operating these services.
Science and Research is depending on the results of their computations. Just with this,
everyone is depending on systems and operating systems used. So problems most imminent
arise especially with the
• Large number of cores,
• Large number of nodes,
• Distributed memory usage,
• Large number of large hard disks,
• Read and write speed of storage.
With the increasing number of requests and interactivity the communication size, size of data,
transfer band width, scalability, and mean times for failure get more important. An intelligent
arrangement and configuration of system components and an overall management of system
components gets into the focus.
The most prominent problemwith the next generation of resources is quantity of components.
The handling of quantity leads -besides many other challenges- to increased demands for
encryption, IO, PCI, on-chip features, error correction (ECC), research and development,
scientific and academic staff and supporting maintenance, operative and administrative staff,
as well as for secondary dependencies like energy resources and unbreakable power supplies.
4.1 Consumption
The most prominent problemwith quantity, besides the computing obstacles, is consumption.
State of the art power and energy measures are for example Low Voltage memory (LV
DIMM), Light Load Efficiency Mode (LLEM), multiple Power Supplies, watercooler chassis
& air conditioning, higher temperature cooling, hot water cooling, hybrid cooling systems,
Energy and Power Manager (Active Energy Manager, AEM and others), application/energy
frequency optimisation, energy reduced low frequency Processors, Power Management, and
Energy Management.
4.2 Shortcommings
Besides that modular, dynamical applications are rare, even in geosciences, shortcommings
regarding application context and how to handle these aspects are obvious:
• Architectures (CPU, GPU, GPGPU, FPGA, . . .),
• Languages (high level languages, CUDA, . . .),
• Memory,
• Fast and broad band Networks,
• Efficiency,
• Manageability, . . .