Tartan High Performance Computing Initiative

Tartan is a multi-year initiative to provide NJIT researchers the centralized computational and data storage resources necessary to support their research. The name "Tartan" (from the concept of weaving together elements of varying colors and widths) was chosen to reflect researchers' varying hardware requirements and funding capabilities. Researchers' high performance computing (HPC) needs vary according to the nature and scope of their computational problems; thus, the HPC infrastructure must support computational hardware of various capacities and technologies, and accommodate a wide spectrum of data storage needs. This model is inherently capable of incorporating new technologies as they become available.

In the very rapidly changing areas of HPC and big data, flexibility and agility in systems design is crucial. Some researchers may need many CPUs (central processing units, or cores) with small RAM, some a few CPUs with large RAM, some a mix of CPUs and GPUs (graphics processing units), some the CPU and storage framework for dealing with very large datasets and big data. Tartan is designed to accommodate such needs, and evolve to accommodate new needs as they arise.

The vision for supporting the computational research needs of our faculty is to provide a foundation of infrastructure so that junior faculty can establish a record of scholarship. Based on that record of scholarship, established researchers will bring external funding to expand the computational infrastructure to meet the needs of their expanded research agendas.

Established researchers' investments in Tartan guarantees them dedicated resources commensurate with their investments. In addition, individual owners of computational hardware that is part of Tartan can choose to share, at periods of their choosing, the use of that hardware with other researchers. Thus, researchers can lend resources to others during periods of low need, and use others' resources in addition to their own during periods of high need.

Tartan is provisioned with the highest quality components, incorporating redundancy and resiliency, and designed to last many years - in general far longer than the commodity components that researchers in the past have purchased on an individual basis. When investing in Tartan, researchers get not only their dedicated hardware, but also all of the support infrastructure for that hardware : datacenter physical environment (power conditioning, cooling, UPS, and physical and top-level network security), high speed networking, enterprise backups, and professional systems administration staff (ensuring that systems software and security are kept current). Inherent in the Tartan design are the options for researchers to readily expand their dedicated resources by the purchase of additional processing power (cores), RAM, and storage, as well as new types of CPUs and/or GPUs; these expansions are generally seamlessly integrated into the researcher's existing Tartan hardware environment. Such options are most often not available for individually-owned hardware. Although in some cases this total package will be slightly more expensive than the basic hardware that researchers can purchase, house, and administer on their own, when all costs are considered the investment in Tartan is the better value.

Tartan Goals

  • Meet all expectations of researchers using HPC :
    • Computational power
    • Software infrastructure
    • High-capacity internal and external networking
    • Support for very large datasets and big data
    • Parallel file storage
    • InfiniBand cluster node interconnects
    • Easy expansion of hardware resources
    • Base allocations of resources in the central shared infrastructure
    • Professionally-managed datacenters providing a proper physical environment
    • Enterprise backup
    • Substantial and useful documentation
    • Training sessions, workshops
  • Support a robust HPC & BD infrastructure :
    • Provide the capability of supporting leading edge research
    • Provide researchers a more desirable option than self-provisioning HPC equipment
    • Maintain the HPC infrastructure at a level consistent with the needs of researchers and the maintenance of a competitive stance relative to NJIT's peers
    • Provide university-supported HPC resources for junior faculty, to support the production of scholarship that will lead directly to external funding

 

Back to the Cyberinfrastructure page

 

Last Updated: March 23, 2017