Tim Brown and Mehran Basiratmand
Monday, August 17, 2015
- Established in 1972 as one of four regional data centers for the State University System of Florida, NWRDC initially offered mainframe services to universities across the state as a community cloud system.
- A self-governance model makes NWRDC a computing cooperative of over 70 member organizations with access to enterprise-level services and facilities that would be difficult and expensive to implement individually.
- Still heavily rooted in education, NWRDC now provides services to a wide range of universities, colleges, and state, county, and city governments.
The essential goal of any cloud service is to use resources efficiently in an “on-demand” model that particularly benefits educational institutions. A community cloud exists within Florida State University’s Northwest Regional Data Center (NWRDC). Established in 1972 as one of four regional data centers for the State University System of Florida, NWRDC initially offered mainframe services to universities across the state.
Mainframes to Data Centers
In the “big iron” days of the mainframe, building a shared computing environment was relatively easy and cost-effective. IT could track users by CPU, I/O, and their use of storage. While some organizations charged back for the time used, others were centrally funded as part of their institution, and their users considered it a “free service.” Mainframes were much too large and expensive for everyone to have one, and the shared model offered a logical alternative. The end-user environment consisted of cheap, dumb terminals.
With the evolution of the microchip and the advent of personal computers, the computing workload became dispersed. Subsequently, client-server computing provided the opportunity for distribution of some tasks to the local systems, thus reserving the central computer for heavy calculation, specialized applications, and data distribution. Another evolutionary step led to web-based applications, where computing is handled by a mixture of back-end servers also known as multi-tier environment. With this step, the workload on the local PC decreased.
Although easy to create, the modular or tiered components (application, database, web server) require multiple servers. Instead of housing many applications on one set of servers, this approach created “system silos” of many applications duplicated across the enterprise. Some of these silos resulted from technical and resource limitations, separating one application from another. On the other hand, most silos arose for managerial or political reasons. While we could share some of the infrastructure components such as enterprise storage, networks, or, in some cases, database servers, this led to “system sprawl,” filling data centers with many small, underused systems.
The next step in systems evolution took a lesson from the past. Mainframes had long been able to create logical partitions within one physical server. The rediscovery of virtualization for use on open system platforms let us use the idle cycles on underused servers. We could now provide isolated environments for applications. Virtualization also let us create generic platforms that we could easily move between different hardware platforms and operating systems alike. This greater portability led to the next “rediscovery,” known as cloud computing. When we connected to the mainframe for services, we did not usually care, or even know, where the mainframe sat. Virtualization (and high-speed networks) gave us the same freedom. It no longer mattered where the compute cycles physically sat with the dawn of cloud technology.
There are four basic types of cloud environments:
- Public. When you hear the term “cloud computing,” usually a public cloud comes to mind. Public clouds can be large, with customers coming from many different areas of business. Large public clouds can be distributed nationally or even globally, involving multiple data centers. In this model, customers might not know which data center serves their data, and some customers may feel concern over their data leaving the country. This model supports a large customer base and maximizes resource sharing; however, given its geographically disperse nature, many higher education organizations have concerns as to the whereabouts of their data, especially HIPAA-compliant or intellectual property data sets.
- Private. Some people argue that private clouds are not really clouds at all, in that they belong to a single organization. While the owner can rearrange resources and direct them where most needed, the organization still has purchased or acquired all of the resources, creating a strong likelihood of underused resources. This means that any resources not currently used are an expense without benefit to the organization. On the positive side, the customer knows where all data is stored, that the risk of access by unauthorized personnel is minimized, and that tighter organizational control could be implemented if needed.
- Community. A community cloud leverages the best features of public and private clouds. Resources are shared, but among organizations with a common function. The customer base may consist of universities within a system or a consortium of school districts. Multiple customers provide for more effective use of resources, and the common focus allows for a greater level of control and security. For example, a community cloud for education entities might put a greater emphasis on the security requirements mandated under FERPA, as well as other stringent requirements sanctioned by federally funded research activities. Furthermore, the community cloud gives a greater level of specialization and control while still sharing resources and costs.
- Hybrid. A hybrid cloud is any combination of the first three models that allows sharing resources between local sites and a community-based cloud. This model recognizes the need for greater availability of and access to resources in a shorter time-frame, and for peak usage customers could augment their local cloud resources by tapping into the community cloud.
Any cloud service aims to use resources efficiently in an “on-demand” model so that the user only pays for the resources used. This type of service should be very attractive to educational organizations that have well-defined peak use periods such as registration, admission, or financial aid cycles. For example, an institution can subscribe to additional computing power for fall term registration, then release the computing resources after that period has passed. You no longer have to pay for extra capacity that sits idle until you need it.
Community Cloud Data Center
An example of a community cloud exists within Florida State University’s Northwest Regional Data Center (NWRDC). Established in 1972 as one of four regional data centers for the State University System of Florida, NWRDC initially offered mainframe services to universities across the state. By 1976, NWRDC was providing services for the Florida Department of Education, which led to additional customers comprised of the various school districts within Florida. While still heavily rooted in education, NWRDC now provides services to a wide range of universities, colleges, and state, county and city governments (see figure 1).
Figure 1. NWRDC customers
A common concern over cloud computing — especially community cloud technology — is loss of control. Many people believe that organizations generally have the best interest of their users in mind; however, can users trust an external group similarly? To address this concern, NWRDC adopted a shared governance model to give all participants a sense of ownership. This move has essentially alleviated the trust issue and been key to the community cloud service's success.
While NWRDC organizationally reports to FSU for operational activities, for all policy and strategic issues (including budget allocation and large-scale purchases) it reports to a Policy Board comprised of its customers. This shared-governance model has been successful for well over 40 years and is detailed in NWRDC’s Charter of Operation. Board seats are based on the amount a customer spends with the organization. For example, a customer whose expenditures make up three percent of NWRDC’s overall revenue receives a voting seat on the Policy Board. Additional seats are granted at different levels of expenditure. To make sure smaller customers are represented, a voting seat is established to represent anyone whose expenditures do not warrant a seat of their own. In recognition of the importance of the K–12 community, they are treated as a single consortium and given a voting seat of their own.
The 11-member Policy Board approves NWRDC's annual operating budget. This unique governance model actually gives customers the authority to approve the rates they will pay. They also approve all new services and customers. This self-governance model makes NWRDC a computing cooperative comprised of over 70 member organizations. By working together, the member-customers have access to enterprise-level services and facilities that would be difficult and expensive to implement individually.
A self-funded auxiliary, NWRDC receives no funding from FSU or the State of Florida — it is 100 percent funded by its charges for services. As a nonprofit state entity, NWRDC cannot charge more for a service than what it actually costs to provide. Each service or cost center must be self-supporting and operate within the guidelines of the Charter of Operation.
When the State of Florida began a data center consolidation effort in 2007, it modeled the organization after NWRDC. In recognition of its long history of customer service, NWRDC was made a consolidation point for other state data centers. This allowed the development and introduction of many new cloud-based services, including storage, application, infrastructure, backup, etc. While the customer base is small compared to large public cloud providers, by working together the NWRDC community cloud members have achieved a pricing structure that is competitive to its public big brothers.
Florida’s Department of Education has been a customer of NWRDC since 1976. Levis Hughes, who leads DOE’s Office of Student Financial Assistance, stated:
“OSFA has had two recent opportunities to utilize NWRDC’s Cloud Infrastructure services. Our file management system recently began using cloud services and allows OSFA to process over 240,000 file transmissions monthly. Our transition was quick and smooth. Additionally, OSFA is in our testing phase of redeveloping our State Scholarship application, which processes over 200,000 student scholarship eligibility applications for high school graduates. This service offering now allows us to reduce floor space costs, reduce potential reconfiguring of systems, reduce need to update software and software upgrade monitoring, and increases our flexibility by starting with a baseline that matches our needs but allows modifications in quick order when needed.”
The shared governance model at NWRDC has facilitated the opportunity to explore and deliver leading-edge technology solutions based on a shared-cost model. Given the enhancements to virtual machine technologies, NWRDC can position itself to go beyond community cloud services and provide a hybrid cloud where services shift to NWRDC on ad-hoc basis. This creates a capacity on-demand model for member-customers, and the future of this service looks truly promising.
Tim Brown is executive director of the Northwest Regional Data Center. He earned his bachelor's of science degree in applied physics from Auburn University, and his master's of science in health informatics from the University of Alabama at Birmingham. In addition, he is a Certified Information Systems Security Professional and Certified Information Systems Auditor.
Mehran Basiratmand is the chief technology officer at Florida Atlantic University and the chair of the Policy Board for the Northwest Regional Data Center. He earned his PhD in higher education leadership at Florida Atlantic University, a master's of science in computing from Barry University in Miami, a certificate in systems dynamics from MIT, and a certificate from the EDUCAUSE Leadership Institute.