In response to the data explosion in higher ed—and the value to be gleaned from it—institutions are rethinking data management and the traditional data center .
Everyone has heard the meme or a close variant: 90 percent of all the data in the world has been created in the last two years. As suspect as such statistics may be, they nevertheless speak to an underlying truth: We are inundated with data as never before.
And for hard-pressed IT professionals at colleges and universities, much of that data seems to be coming out of their institution alone.
In higher education, various factors have conspired to form a perfect storm of data generation: The migration of administrative and instructional functions online, sensor-driven research, supercomputing, and, most recently, the Internet of Things. How do schools store all this data while ensuring its security and recoverability? And, in an era of consumer IT, how can they guarantee that this data is available upon request?
These considerations—as well as cost—are the driving forces behind the three biggest trends in data center management in higher education today:
(Next page: Outsourcing and modularity)
1. Outsourcing
Universities live on data; it’s central to everything they do, so they are understandably wary about entrusting it to a third party offsite.
“The issue of ceding control is tricky,” said Peter Ritz, CEO and co-founder of Keystone NAP, a third-party multi-tenant data center in Pennsylvania that serves institutions and companies in the Northeast. “Administrators ask themselves, ‘Do I want to rely on someone else for something that is so mission critical?'”
At the same time, changing expectations among university constituents are forcing schools to rethink their approach to data storage. “As a university, everyone depends on your systems being up and running,” explained Ritz. “The consumerization of IT has set the bar pretty high for everyone to drive high service levels, and universities are finding it difficult to attain those levels on their own.”
For data centers, the keys to high service levels lie in 24/7 availability and connectivity, guaranteed via built-in redundancy. As schools in the Northeast discovered during Hurricane Sandy, for example, loss of electricity at the data center—needed to power the servers and provide cooling—can effectively shutter a campus.
“Even though learning happens in a classroom, the rest of it happens online: the research, the homework, and everything else,” said Ritz, who noted that most universities have no more than one or two links to the electrical grid, whereas Keystone NAP has five. “If you’re not available, resilient, and reliable from an online perspective, it’s very hard to run a university.”
Unfortunately, building the kind of scalable infrastructure and multiple redundancy needed to ensure 24/7 uptime is an expensive proposition. Not surprisingly, many schools are deciding that building and running their own data center is neither cost-effective nor part of their core competency.
“Data centers, especially at schools doing heavy computational workloads, require a kind of discipline that universities find non-core,” said Ritz, adding that it’s seven or eight times more likely that a customer today will contract with a co-location provider than build its own data center. “When you look at the kind of capital expense and commitment needed, it all makes sense. Many schools find it much more productive to invest in more sophisticated labs or research facilities.”
2. Modularity
Modularity has been around for several years, but it’s gaining traction as new, more reliable configurations come on market. The concept is simple: Customers maintain their servers in their own self-contained storage units which are then plugged into an established infrastructure—described as a “chassis” by Ritz—which provides redundant power, cooling, and connectivity.
Need more storage? Simply plug in another unit. This approach can provide a trifecta of benefits over old-style university data centers: Improved data security, greater reliability, and rapid scalability.
A modular data center is only as good as the underlying chassis, however. For starters, the physical site must be secure, protected against flood, fire, earthquake, and intruders. Second, it must provide multiple connections to independent power sources. Third, it must be well networked, with access to redundant connections to major fiber routes.
For its modular storage units, Keystone NAP utilizes what it calls KeyBlocks, manufactured by Schneider Electric. The modules come in 22- or 44-rack increments, with the smaller of the two units occupying approximately 1,000 square feet. “The benefits really are in the ability to deploy very quickly,” explained Ritz. “If you tell somebody you can get 1 megawatt to 20 megawatts’ worth of workloads staged in three months, people don’t believe you can do that.”
As for what happens in a university’s storage units—and who has access to the data, whether physically or remotely—it’s completely up to the school.
“We think about it in terms of a private vault at a bank,” said Ritz. “The security equipment only authenticates the people who are authorized to be there. It’s not us unless the university has contracted with us for managed services.”
(Next page: The promise of the Hybrid Cloud)
3) Hybrid Cloud
After years of hype, hybrid clouds are beginning to live up to their early billing.
A hybrid cloud generally refers to a combination of a private cloud—an onsite or third-party data center, for example—and a public cloud, such as Amazon Web Services (AWS) or Microsoft Azure. Until recently, hybrid clouds were limited by connectivity issues between the two types of cloud, but cloud providers have largely overcome this problem by offering private network connections that bypass the internet altogether. AWS, for example, offers a service known as Direct Connect.
A hybrid cloud approach is likely to appeal to universities that are leery of storing data in a public cloud but nevertheless want to take advantage of the computing power of cloud services. A good example is the student data stored in a school’s CRM, said Ritz: “Imagine the university wants to hire a firm to run some analytics on CRM data to identify trends. It could push this data through AWS Direct Connect, spin up a bunch of AWS servers in Hadoop clusters, run those workloads there, and then bring the information back.”
In effect, a hybrid cloud gives schools a way to unlock the value tucked away in its data center without surrendering the keys to the castle. “When it comes to data, you want to have compute right next to it,” noted Ritz. “Creating a network link between the data center and a public cloud gives you a shortcut to get there quicker.”
Andrew Barbour is an editorial freelancer with eCampus News.
- Guidance on higher ed revenue sharing may be safer than many assume - December 6, 2024
- Students using AI: It’s not that scary and shouldn’t be banned - December 2, 2024
- Identity theft preys on campuses–here’s what we can do about it - November 26, 2024