The On-Site Data Center and Off-Site Cloud Solution: Blending the 'Private Cloud' with the 'Hybrid Cloud'
Deciding between whether to build a new data center on campus to replace an outdated 40 year old facility or move the entire operation into the cloud proved a most challenging dilemma for this CIO who strives to be both visionary and pragmatic.
My role as visionary CIO, flying at 30,000 feet, had all the fancy strategic answers: The cloud is the only option for a future-focused institution. Clearly, handing-off this infrastructure to the “real experts” would better position us to support our core business while also ensuring that we have highly resilient systems. Finally, we would become a more agile operation, quickly responding to our campus customers’ needs—perhaps at a much lower cost!
My role as pragmatic CIO, operating at ground-level most days, had all the gloomy scenarios that leaping to the cloud might create. Beyond the uncertainty that change of this magnitude brings, I was very worried about relocating and supporting the 400 servers we currently host for numerous campus clients. Where would we find the staff to manage this cloud transition while maintaining current operations? What impact would “off-siting” the data center have on our Internet bandwidth? And perhaps most importantly, how would we fund and manage a migration of this size and complexity?
Given that our existing data center had proved to be inadequate for an institution of our size, we first determined whether or not it was “fixable.” Outside experts were hired to assess the current facility and provide recommendations. The answer was clear: don’t spend money fixing the old data center—it’s too costly and not sustainable for the long term. With that recommendation in hand, we considered two options: 1) Build a new data center in available space of another building or 2) Identify and migrate to a co-location facility. We engaged in this review in 2008—the early days of cloud-based infrastructure. At that time, co-location generally involved placing your equipment in a leased/caged space in a commercial data center.
As you might suspect, campus leadership was a bit uneasy asking the IT shop to conduct an assessment without outside validation. We utilized two different external consultants: the Gartner Group provided us with a validated framework for assessment metrics and Lexis-Nexis, a nearby multinational, arranged for a “loaned executive” to validate our “co-location versus build” analysis. In this case, the data strongly supported building over co-location. Our new 5,000 sq. ft. production data center (pictured here) was completed in 2012. It provides us with the ability to rapidly respond to planned growth in both enrollment and research volume. The co-location sites considered in 2009 were not able to cost-effectively guarantee anticipated expansion requirements.
The University of Dayton’s new data center not only established a “private cloud” with an extensively scalable virtualized capacity but also showed that our engagement with the external cloud blends well with this on-site facility. In addition to supporting over 150 physical and 300 virtual servers with nearly a petabyte of storage in our data center, we have also implemented nearly 40 applications and services in the external cloud. Changes in cloud services over the last five years have made the benefits of moving to the external cloud much more significant for higher education entities. Co-location is no longer the primary “off-site” option. Today, cloud services fall into three distinct categories:
Types of Clouds:
1. IaaS: Infrastructure as a Service—the Amazon/Rackspace approaches whereby you remotely deploy your systems on provisioned processing, storage or network capabilities.
2. PaaS: Platform as a Service—the Microsoft Azure approach whereby you operate your services on preconfigured and scalable operating systems and applications.
3. SaaS: Software as a Service—the “all in approach” whereby you utilize a vendor’s software application on their remotely-based and vendor-managed applications and servers.
While different organizations see value in each of these options, we have found that “Software as a Service” fits wonderfully with our current resources and staffing levels. As with most institutions of higher education, the growth of new applications and related services is explosive and few IT organizations have the capacity to continually deploy and sustain the growing number of software packages. While we have the physical capacity to add more services, we don’t have the staffing to support every application that is adopted. Thus, we have developed a very effective and comprehensive model for evaluating a new system and determining whether or not it should be managed on-site or in the cloud.
“Changes in cloud services over the last five years have made the benefits of moving to the external cloud much more significant for higher education entities”
By way of definition, the “Hybrid Cloud” is most often characterized as multiple cloud systems that allow for easy movement of programs and data across implementations. “Software as a Service” typically falls into the category of the “Hybrid Cloud” because the applications are distributed across multiple cloud providers with interfaces that frequently connect with your private cloud for data sharing, authentication or other similar services.
Moving to the cloud has varying levels of challenge, depending upon integration complexity. The examples given in the Table show the typical integration complexity we faced with our cloud-based applications. Storage/Access/Content tend to be least difficult to deploy and sustain within our enterprise infrastructure. Tools supporting communications raise the level of complexity, but not extensively unless you are beginning to scale to the entire enterprise (such as moving to Google Apps). Finally, administrative applications—those requiring authentication, data exchange and integration with other hosted or on-site systems—always demand greater effort and oversight.
In the final analysis, private clouds provide the ability to support complexity at home while strategically enabling careful and thoughtful migration of selected applications to cloud. While we have not found substantial cost-savings by deploying applications to the cloud, we have discovered that clouds do provide the following advantages:
1. Staffing Availability & Skillsets:SaaS overcomes common IT staffing limitations.
2. Agility: With some applications, flexibility is the key to early success.
3. Time-to-Market: Speed to market frequently makes a big difference for units such as enrollment management or marketing.
4. Innovation: Being on the cutting edge can truly differentiate institutions in highly competitive markets.
With frequent media hype about “all things cloud” being the future of “most things IT,” it’s difficult to loudly sing the praises of a modest cloud strategy while wearing your “visionary CIO hat.” I have no doubt that one day in the future we will think of the IT cloud much like we think of most utility services—a “no brainer.” Who would suggest that we build a water system on campus when reliable and clean water is so widely available from public sources? For today, unfortunately, not all IT applications can easily, quickly or cheaply be moved to the cloud. This is particularly challenging with limited staff, uncertain funding and the expectation that we continue all existing services with “24x7 up-time.” At the end of the day, I believe moving to the cloud via a hybrid model with a deliberate migration strategy is the best solution for CIOs who aspire to become “pragmatic visionaries.”