Hyperconvergence hits higher ed IT
Hyperconverged infrastructure (HCI) has been a mainstay in the corporate world for a few years, and the wave of higher ed IT leaders now moving to this model say the buzz around it is warranted.
HCI incorporates computing, storage, networking and virtualization resources. This integration reduces the amount of hardware needed, changing the face of data centers.
The technology is gaining momentum. Sales of hyperconverged systems surpassed $12.5 billion in 2017, representing a 9.4 percent increase over 2016, and it was the first time the market surpassed $12 billion in a calendar year, according to International Data Corp. (IDC), an IT market intelligence and advisory services provider.
IDC’s analysts note that what differentiates hyperconverged solutions from other integrated systems “is their scale-out architecture and their ability to provide all compute and storage functions through the same x86 server-based resources.”
As campus IT leaders look to replace aging infrastructure, they say they want to invest in technologies that provide ease of management, a reduction in administrative overhead and a return on investment.
Planning for the future
Hyperconvergence appealed to Loras College in Iowa when leaders began replacing aging equipment because one vendor would provide all networking components, says Tom Kruse, CIO and director of budgeting and procurement. Declining enrollment at the college, which currently has about 1,450 students, also influenced the decision.
“We’re down about two full-time equivalent staff members [in IT] and that’s where HCI comes in,” Kruse says. “It’s much easier to administer.” And with computing, storage and networking on one platform, he adds, “it’s a one-stop shop.”
The specter of college mergers also loomed over the HCI decision, which became a no-brainer, Kruse says, because HCI allowed him to scale up his three servers to 10—all linked by virtual machines. Loras’ leaders are not currently considering a merger, Kruse points out, but other institutions have begun sharing IT and other services without merging. The college selected Lenovo’s HCI offering.
Mariusz Nowak, director of infrastructure services at Oakland University in Michigan, shares the enthusiasm for hyperconvergence.
“You talk to vendors and they’re happy to sell you back-end storage, which is a half-million dollars, and then you pay hefty prices every year for support, maintenance and patches to make sure that the storage works well,’’ he says. Adopting HCI is a way out of that cycle.
Nowak began looking at hyperconvergence four or five years ago when enrollment at Oakland was projected to decline due to decreasing numbers of high school graduates.
As a state school, “we’re pushed to find more efficient ways to recruit students to the university,” he says. That prompted a need for technology that would improve processes and procedures.
HCI, provided by VMware, brought a lot of advantages, including the economics. Nowak liked that storage, memory and compute came together in one package, and he was able to pay for the initial installation with one-time funding. “Now, if I want to grow—and if I don’t get extra money—I’m able to add extra servers, modules and nodes, and grow storage, and retire old servers because HCI is modular,” he says.
Better availability and uptime
The decision to move to HCI was less about declining enrollment and more about improved availability at the Wharton School of the University of Pennsylvania, which is using Nutanix.
“We had a consistent problem of outages—significant ones that impeded research—about once a year,” says Matthew Frew, IT director for enterprise services and solutions. About three years ago when IT was doing a platform refresh, Frew looked at HCI, but “it wasn’t where we wanted to be,’’ he says. The components were still separate, so there was no ease of management.
Frew also wanted to eliminate hurdles to innovation such as incompatibility between firmware and drivers, and having to perform frequent manual updates.
Hyperconvergence now checks off many boxes: offering low administrative overhead, automated updates, and proactive monitoring and alerts to ensure that the platform is stable and can deliver these services reliably.
Hyped up: Hyperconvergence 101
Hyperconverged infrastructure (HCI) is ideal for higher education, which typically has small IT teams tasked with serving multiple stakeholders, says Gil Haberman, director of product marketing at the provider Nutanix, which offers a Hyperconvergence Toolkit for Education IT (UBmag.me/nutanix). The main benefits of HCI are:
Ease of deployment: Hyperconverged software can run on commodity hardware all in one package. In one hour, IT can quickly deploy compute, storage and networking resources, Haberman says. This is useful when IT needs to create a network in a remote or branch office, or a development environment to test a new app. This translates to happier users because they get their requests fulfilled faster.
Management simplicity: Software and hardware upgrades can be done with one click across the stack, improving operational efficiency and infrastructure availability. Typically, an admin has to do manual upgrades and ensure compatibility among packages, increasing the possibility that something will break. This can lead to downtime and unhappy users.
Scalability: An HCI node includes all the necessary elements, and more nodes can be added easily. This helps avoid over-provisioning.
Cost: Fewer resources needed for day-to-day management and more efficient resource utilization are attractive. There is also less downtime, and less hardware, which saves energy, time and money.
“We want to say, ‘The environment is running, it’s up to date, it’s secure, and it takes a few cycles to maintain the operations of the virtual environment, and we can run services on it in a supportable and stable manner.’”
In 2016, Frew attended VMware’s VMworld, a global conference focused on virtualization and the cloud, and discovered hyperconvergence had come a long way.
“That triggered an aha moment for me,’’ he recalls.
Other hyperconvergence considerations
Implementation at Loras was easy, says Kruse, who used a third party to help migrate data off old systems. HCI has been running at the college for about 2 1/2 years now, he says, and “we have hardly had any issues.”
Now, because everything is part of one system, vendors can’t play the blame game if something goes wrong. “Back in the day, we’d have outages and get called after hours, and now, my network guys don’t get called about them any longer,” he says. “They don’t work overtime anymore.”
That said, HCI may not be the best approach for an IT shop that just bought a new storage area network (SAN), Kruse notes. It was an easier decision for Loras because the college’s SAN was 8 or 9 years old and beginning to fail.
“There are a lot of SANs today that probably cost more than what we put into our entire hyperconverged system,” he says, adding that less hardware also means a reduction in energy costs.
The biggest challenge for Wharton’s IT group was “not changing our mindsets and assuming our legacy processes were the way we should do things in the new platform,’’ says Frew. For anyone considering deploying HCI, he suggests first speaking with a vendor’s support department for clarification on how it all works.
Frew encourages other IT teams to change their mindsets and make the leap.
“Administrators spend a lot more time running a three-tier architecture,’’ he says. “Their time is spent far more efficiently running the applications than the platform.’’
While Frew knows what his capacity levels are today, he says he could receive a request tomorrow for compute capacity for a storage-heavy workload, given the push toward gamification in the classroom, for example. Now, he says, it’s much easier to apply another node to scale up resources fairly quickly.
“The world is going more and more ‘cloud first’ and a hyperconverged solution is a great way to balance the two.”
Esther Shein is a Framingham, Massachusetts-based writer whose focus is technology.