Facilities - IT - 2011-2012

IT Mission

To provide an innovative and robust compute environment which facilitates research, fosters collaboration among research groups, and assists in the technical training of students in Earth sciences. Maintain a secure and reliable environment while providing remote access and flexible solutions to diverse needs. Leverage modern technologies to provide a scalable, economical and sustainable high performance environment.

General Information

The Institute's computational facility is a reflection of the features of the research unit itself: a unique, shared, and communal resource allowing interdisciplinary and collaborative research and training to flourish. The open nature of the shared computational resources is unprecedented in U.S. research groups. Most importantly, the community computer resource enables students and faculty researchers to share not only hardware and software resources but also the data sets and specialized computer programs that are the core of the individual research projects. This sharing of intellectual achievements enables institute researchers to easily make new and important Earth system science and integrated assessment discoveries while sharing results quickly with the wider community, which provides a truly interdisciplinary environment in which to train students. The Institute supports:

  • ≈150 UNIX systems
  • ≈150 Macs
  • ≈50 Windows PCs
  • 4 Windows Servers
  • 13 managed network switches
  • 17 networked printers
  • 4 Linux HPC Clusters with a combined total of ≈104 nodes
  • ≈500 Tb of disk storage

Networking

The research unit has 1000Mb/s connection from the UCSB campus backbone to server rooms in Girvetz and Ellison Hall. This provides shared access to a 622Mb/s CALREN-2 connection, which in turn provides access to Internet2. High speed layer two switches and WAPs provide Ethernet, Fast-Ethernet, Gigabit-Ethernet and Wi-Fi connectivity. The Institute's network spans 3 class C subnets and extends to several campus locations via VLAN tagging and additional small subnet allocations across the campus backbone. Locations include: Ellison Hall, Webb Hall, Girvetz Hall, Bren and Harder Stadium. Environment The computing environment is based on a network of primarily Linux-based (x86) hardware but also includes Sun Microsystems (SPARC and x86), and HP Compaq Digital (Alpha) servers and workstations. The computing environment's architecture is designed to permit rapid deployment and easy integration of new hardware. Virtual systems based on the open source KVM and Xen projects are also available to institute researchers providing rapid, inexpensive, flexible and reliable resources. Vast datasets of MODIS, TM, AVHRR, to name just a few, are readily available on-line to all researchers at the Institute as are the tools and software for modeling and other modalities of scientific analysis.

High Performance Computing

Two of the primary Linux Clusters consist of 22 AMD 2800+ MP CPUs, 22GB of RAM and 2TB of dedicated, high-speed disk space and the other of 35 AMD 2500+ CPUS, 35GB of RAM, and 8TB of dedicated, high-speed disk space. The clusters are built with the flexibility to add more resources quickly and easily should participants' needs change. IT staff also manages the "Dragon" cluster; a 32 node (64 1.8Ghz AMD Opteron CPUs) compute cluster owned by Professor Chen Ji. The cluster has Gnu and IBM compilers linked in to both LAM-MPI and MPICH MPI configurations. The head node has 9Tb of disk space and 8Gb of RAM. Each compute node has 4Gb of RAM.

Desktop Computing

Windows and Mac systems predominate on desktops which integrate with the general compute environment. High-performance and inexpensive SATA based RAID disk arrays allow participants to add disk storage to the environment in reliable, disk-sized discrete increments. Nightly backups to off-site RAID arrays minimizes the risk of critical data loss. There are ten networked printers including two color laser printers. Finally, a full compliment of computational, image processing, statistical, database, graphical, scientific visualization, and animation software are available for use by our researchers.