Emerging Cyberinfrastructure
Cyberinfrastructure
refers to the rapidly evolving software, hardware, and human expertise
required to support current and future discoveries in science. Just as
conventional infrastructures such as roads, bridges, power grids, and
phone lines support modern society, the cyberinfrastructure (which
includes computer, information, and network technologies) provides
long-term platforms to empower scientific research.
CISL researches diverse components of the emerging cyberinfrastructure, working with experimental systems that promise to increase computing power, data storage capacities, and network performance. CISL is also actively engaged in developing new knowledge environments, an area that merges with CISL's extensive R&D in data and knowledge systems.
CISL evaluates leading-edge technologies on an ongoing basis in its quest to maintain the most cost-effective and efficient computational environment for the NCAR/UCAR scientific community. CISL is currently exploring the possibility of a major expansion of supercomputing capabilities dedicated to Earth systems research. This expansion would capitalize on NCAR's more than three-decade position as the only major supercomputing center within the geosciences community, and one that provides an integrated set of research, services, and outreach.
Experimental systems
Rapid changes in scientific needs and computer technologies mean that UCAR's computational environment must evolve and stay flexible. CISL tracks emerging technologies, assesses capabilities of new systems being offered by vendors, and evaluates programming environments. CISL's work with experimental systems ensures the selection of the most appropriate computers for scientific simulation.
Examples of experimental systems that have been evaluated since the 1990s have included the Meiko Computing Surface, the Connection Machines 2 and 5, various microprocessor clusters, a number of Linux systems, and IBM's BlueGene/L, which in 2005 was rated the fastest machine in the world.
Distributed file systems
In the early days of supercomputing, mainframe computers were shared by all the users in an organization. Nowadays, however, computing environments consist of distributed workstations, servers, and supercomputers, necessitating file transfer via networks and across different platforms and operating systems.
CISL is exploring a number of new technologies for managing distributed file systems, including a Storage Area Network for sharing datasets between data servers and over the Internet, and a high-performance Parallel Distributed File System for sharing files between supercomputers.
Networking
CISL is constantly pursuing ways to provide scalable, sustainable high-performance networking to support the geosciences community. CISL is a principal collaborator in a number of nationally recognized R&D networking and data communications technology projects, including the National LambdaRail and Teragrid.
CISL was a key player in the recently completed, NSF-funded Web100 project, which developed tools to help end-hosts automatically achieve high data rates. CISL also contributed to Net100, a project to create software that allows computer operating systems to adjust to available network bandwith for large data flows.
Other CISL network research projects include the Network Path and Applications Diagnosis (NPAD) project, aimed at improving tools to diagnose network performance, and the Hybrid Optical and Packet Infrastructure (HOPI) testbed, which will provide a facility for experimenting with future network infrastructures leading to Internet2.
