Cyber Infrastructure Plan 2016-2021

Florida Atlantic University Cyber Infrastructure Plan 2016–2021

Executive Summary

As part of the strategy to make Florida Atlantic University a top-tier research university, the Office of Information Technology strives to provide the experience and technology to support cutting-edge education and innovative scientific research. FAU plans to enhance and build the IT infrastructure across numerous campuses in support of these goals.

Infrastructure Highlights

  • Dedicated Research Computing Data Center on the Jupiter Campus
  • Green Certified Data Center in the College of Engineering
  • 10 Gb/s Link to Florida Lambda Rail in Boca Raton
  • 10 Gb/s Link to Florida Lambda Rail in Jupiter
  • FAU’s High Performance Computing Cluster Named KoKo Consisting of 2440 Cores and 18 Intel Xeon Phi Accelerators (as of this reporting)
  • Full Membership in the Sunshine State Education & Research Computing Alliance (SSERCA) since February 2014
  • Full Membership in InCommon since Spring 2012

Services and Expertise

Florida Atlantic University is committed to providing well-defined computational and storage offerings, technical expertise, and support services. Through its unified cluster strategy for compute, BIGDATA, and storage, FAU is able to provide diverse resources to the various departments supporting numerous research domains across the University.

Collaborative Framework

Florida Atlantic University, along with the members of SSERCA, is building a framework that allows researchers at one school to take advantage of the expertise and resources at other member institutions. This collaborative framework consisting of compute and storage has successfully been deployed across three universities and continues to grow to other SSERCA member institutions.

Background

In May 2012 FAU began investigating the possibility of building a Central Research Computing Department with the expressed goal of supporting faculty and student research. Supported by a team in the Enterprise Systems group in the Office of Information Technology, the HPC group is charged with investigating faculty requirements and implementing systems designed to support faculty and students. The group’s activities and priorities are set by the High Performance Computing Faculty Advisory Committee whose members meet monthly. The committee’s current goals include introducing or expanding computational science to a broader segment of FAU faculty, students, and researchers.

Goals for Campus Infrastructure

Information Technology has the ability to empower scientific research and education. Recent innovations in computational science have accelerated the need for researchers and students to have access to quality computational environments such as, but not limited to, HPC, HTC, or BIGDATA. Employees entering the workforce in the next decade will need to process data in parallel and must know how to handle such data on high performance networks. Florida Atlantic University’s mission is to prepare students to utilize these new technologies and to provide a safe, user-friendly environment that empowers them to do so.

The FAU administration has supported the notion of central computational computing technologies and has sanctioned the installation of the KoKo HPC, OpenNebula, and, in the future, BIGDATA, with the installation of HTC and BIGDATA technologies as a component of the KoKo cluster.

The foundation of a strong cyber environment is a group of coherent and well-managed resources. The HPC team at Florida Atlantic University has determined that these resources fall into the following categories:

  • Compute capacity and capability
  • Storage capacity protected from disaster and unauthorized access
  • Network capacity for data sharing data large and small
  • Access to expertise, advice, and training for each resource

 

Infrastructure

Data Center

The Office of Information Technology currently supports four data centers, located in Boca Raton, Jupiter, Tallahassee, and Davie. These data centers provide central support for university activities such as email, web, and other associated services. High Performance Computing has resources in two data centers. The private cloud and academic computing servers are located in the 1900-square-foot data center on the Boca Raton campus. The KoKo HPC is located in a dedicated data center on the Jupiter campus. It is connected to the Jupiter network core at 10 Gb/s and has access to Florida Lambda Rail (FLR) and the other campuses using a 10 Gb/s network link.

Network Capacity

Florida Atlantic University is connected to the Internet via the FLR by way of a 10 Gb/s line at the Boca campus and a 10 Gb/s line at the Jupiter campus. The Davie and Fort Lauderdale campuses are linked together using a virtual cloud switch at 1 Gb/s. The Office of Information Technology routinely investigates ways to improve the network bandwidth to the Internet and between campuses. The aim, and necessity, is to increase these to 100 Gb/s and 10 Gb/s, respectively, over the next 5 years because a number of projects cannot be competitively developed without addressing these limits.

It is crucial that the University increase the capacity between the Jupiter and Boca campuses with 10 Gb/s networking and possibly add an additional 10 Gb/s network link at the Jupiter campus for improved LAN activities. This would improve network failover and resiliency and allow the computational resources at the Jupiter campus to take advantage of improved network performance.

Compute and Storage Capacity

Many current and future projects at FAU require the storage of large amounts of data and compute capacity to process the data. Over the past few years it has been increasingly challenging for individual researchers and departments to provide the required storage and computational resource research needs. The complexity of the software and hardware and the cost of housing such systems and on-going maintenance make a shared infrastructure approach more cost effective for the University and researchers. By seeking support from the subject-matter experts in the OIT HPC group, faculty members and departments are better able to focus their resources on research and academics.

During the first half of the academic year 2016–2017, the resources operated by the HPC group included approximately 2440 cores and 960TB of storage. During the second half of this period FAU plans to add an additional 6,500 cores and 1 PB of storage.

The University is committed to continuing growth of the HPC to a total of 6,400 cores and over 2 PB of storage while remaining at the current Jupiter data center. Pending availability of space and network infrastructure improvements HPC computing hardware may then expand to the second Jupiter data center and/or to the Boca campus.

Integration with Statewide Infrastructure

Florida Atlantic University is working collaboratively with other Florida research institutions to build a framework for sharing data and compute cycles across the state. The University is an active member of the Sunshine State Education & Research Alliance (SSERCA) whose members include Florida State University, the University of Florida, the University of Central Florida, the University of Miami, and the University of South Florida. The alliance holds quarterly meetings and hosts a booth at the annual Super Computing Conference. SSERCA allows FAU to build on the knowledge of other Florida Institutions and access to a wide array of compute resources.

Implementation of Xsede and InCommon

Florida Atlantic University acknowledges that to collaborate with other institutions FAU must offer a federated authentication system. Through our partnership with InCommon, and FAU’s implementation of CAS and Shibboleth, the University is able to bridge its campus-wide authentication system with CiLogon, Xsede, Globus, Research.gov, and many other third parties. The HPC group is currently working to add compatibility with XSEDE resources and implementing assistive technologies that enable researchers to share data quickly with other sites.