Purdue University - Department of Statistics - Computing Facilities Skip to main content

Computing Facilities

Computing facilities and support are provided by the Statistics Department, the College of Science (Science IT), and the central university computing group, known as Information Technology at Purdue (ITaP). ITaP centrally supports many computing resources, including instructional labs, classroom technologies, core network infrastructure, research computing, etc.  The Department, the Science IT group, and ITaP work closely together to provide a robust computing environment that facilitates world-class instruction and research.

IT Staff

Doug Crabill

The Department of Statistics is part of the College of Science, which has a 24-member centralized IT organization to serve the College and Departmental computational needs. The Department of Statistics also employs a Senior Academic IT Specialist, whose role is to provide direct research computing support to the faculty and graduate students. 

Departmental Facilities

Rack Back

The Department of Statistics, with the assistance of the College of Science, maintains around 35 Linux servers and workstations (physical or virtualized) for research and administration. The Linux servers run the Ubuntu operating system. The departmental research servers range in power, but all have between eight and 28 CPU cores. This includes two servers with 1TB RAM, one with 768GB RAM, three with 512GB RAM, one with 384GB RAM, three with 256GB RAM, and others with various smaller memory configurations. These servers have various storage configurations ranging from 168TB on the largest server, to 12TB SSD on the fastest server, to as little as 1TB on the smallest. Total storage capacity across all departmental Linux servers is over 600TB. One of the servers has four Nvidia Titan X Pascal GPUs. Six 64GB servers are part of a Hadoop cluster utilizing 144TB of internal disk space. Most servers with large amounts of storage use a ZFS filesystem with compression for high performance, capacity, and reliability. Smaller servers generally have 1TB to 2TB disk space each, often with RAID mirroring. Some Linux servers use the department's 92TB 10GigE based iSCSI Storage Area Network (SAN) for storage. There are a number of administrative servers providing services like CIFS, configuration management, MySQL, Web, Wiki, printing, backups, etc.

The department, with the assistance of the College of Science, maintains a Windows 2012 domain using Active Directory. It includes multiple domain controllers, redundant print servers, redundant home directory fileservers connected via 10GigE to the 92TB iSCSI SAN, and other administrative servers. Six virtualization servers with 256GB RAM each run Microsoft Hyper-V with RemoteFX to support a VDI infrastructure providing 30 virtual desktops in 8GB RAM and 32GB RAM configurations. They also host many virtual Windows and Linux servers for departmental research, teaching, and infrastructure support. The virtualization servers utilize the departmental 92TB SAN for storage as well. There are 85 Windows desktop computers in offices and computer labs.

Major application software available includes R, Python, SAS, SPSS, Minitab, Matlab, Maple,

University Facilities

Racks

The central Purdue University IT organization, Information Technology at Purdue (ITaP), provides a wide array of computing resources to researchers. Resources include four supercomputers, each with 1-3PB of dedicated scratch storage, more than 15 petabytes of automated tape storage, and a 3 petabyte on-demand storage facility. Campus buildings have at least 20 Gbps connectivity to other campus buildings. There is a 200 Gbps link to the national research network infrastructure. Every year or so the oldest  supercomputer is retired and a new one is built to replace it. The new supercomputers often debut in the top500.org list of the fastest supercomputers in the world. 

ITaP's Rosen Center for Advanced Computing (RCAC) provides a number of research computing solutions.  The Community Cluster Program pools funds from grants, faculty startup packages, and institutional sources to fund supercomputers and make more computing power available than faculty could afford individually. The Rosen Center installs, administers, and maintains those systems so researchers can concentrate on doing research without having to manage a high-performance computing system themselves.

The Rosen Center provides a broad range of technical support services for researchers. These include documentation, training, consulting on system use, software installation, program design and performance engineering, parallel programming, large-scale data management, capacity planning, and cluster deployment.

Computational Resources:

  • Bell Cluster: This cluster has 456 total nodes with nodes having a pair of 64-core AMD Epyc 7662 "Rome" processors (128 cores per node), for a total of 58,368 cores across the cluster.  Each node has 256GB of RAM and has a 100 Gbps HDR Infiniband interconnect. The cluster has a dedicated 3.5PB of scratch space and access to the 3PB Data Depot storage system.
  • Brown Cluster: This cluster has 550 total nodes with node each using a pair of Xeon Gold "Sky Lake" processors with 12 cores per processor, for a total of 13,200 cores across the cluster.  Each node has 96GB of RAM and has a 100 Gbps EDR Infiniband interconnect. The cluster has a dedicated 3.4PB of scratch space and access to the 3PB Data Depot storage system.
  • Gilbreth Cluster: This GPU cluster consists of 102 Nvidia V100 and P100 GPUs in 49 nodes, with each node having between 192GB and 768GB of RAM.   It has 2.3PB of dedicated scratch space and access to the 3PB Data Depot storage system.
  • Halstead Cluster: This cluster has 508 total nodes with each node using a pair of Xeon E5 processors with 10 cores per processor, for a total of 10,160 cores across the cluster.  Each node as 128GB of RAM and has a 100 Gbps EDR Infiniband interconnect. The cluster has a dedicated 2.3PB of scratch space and access to the 3PB Data Depot storage system.
  • Fortress: The Fortress system is a large, long-term, multi-tiered file caching and storage system utilizing both online disk and robotic tape drives.  It utilizes an IBM T3584 robotic tape library with a capacity of over 15 PB. Fortress is a good means to store archived data - files and datasets that will be infrequently accessed by must be retained reliably. Fortress is directly accessible from all RCAC systems, as well as from several other major networks around Purdue's campus

Purdue Department of Statistics, 150 N. University St, West Lafayette, IN 47907

Phone: (765) 494-6030, Fax: (765) 494-0558

© 2021 Purdue University | An equal access/equal opportunity university | Copyright Complaints

Trouble with this page? Disability-related accessibility issue? Please contact the College of Science.