Everybody’s talking about hyperconverged data centers these days, said Jon Junell, assistant director of enterprise infrastructure services at Western Washington University. But WWU beat them to the punch.
“We’ve been hyperconverged for a year and a half, suckers! You end up going down the hyperconverged path because it makes sense for limited staff budgets, reducing the amount of power, and the amount of [data center] space,” Junell said in an interview with SDxCentral.
Nutanix is the vendor going down this hyperconverged infrastructure (HCI) path with Western Washington.
The Bellingham, Washington, university enrolls about 15,000 students. When Junell arrived four years ago, WWU was running legacy Hewlett Packard Enterprise (HPE) EVA storage area network (SAN) arrays connected by legacy fiber. The SAN arrays were slow and required a lot of maintenance, and the service contracts were about to run out.
“We said, ‘Let’s fix the SAN problem,’” Junell said, adding that the university looked at storage systems from HPE, Nimble, and Tintri. “The staff said this is all great, but in a year and a half we’re going to need to replace all of our compute. And we’re not gaining any new people [in IT].”
Around the same time, a small student-sponsored virtual desktop infrastructure (VDI) project was looking for a “dense, relatively inexpensive solution to pilot VDI,” Junell said. This project deployed a small VDI Nutanix cluster in 2014. After seeing success in this case, the IT department started considering Nutanix for its data center upgrade.
“Once we got over the 'let’s buy a storage array from here and a compute network from here,' we moved on to 'hey, we’ve got a great relationship with Nutanix, let’s work with them to move the good stuff,'” Junell said.
Space, Monetary SavingsNutanix costs less, too, he added. WWU was looking at two new HPE storage systems that cost about $650,000 each. “And that was just for storage — that wasn’t including the SAN fabric. For about that same price we got both our compute and storage done and we were able to buy new [Cisco] switches.”
The university deployed a Nutanix cluster in 2015 and another cluster a year later, totaling 12 nodes, or six per cluster. The clusters combine compute, storage, networking, and virtualization into one appliance.
“The space savings is huge,” Junell said. The only downside: 'the smaller data center footprint doesn’t impress visitors,” Junell jokes.
“I take people into the data center and it’s just 6 [units] of stuff. It’s so small and so dense and so efficient, it’s kind of depressing to show people,” he said.
An additional benefit is that the Nutanix clusters use a quarter of the power compared with the legacy system. They also eliminated 20 hours per week of storage and server management time, Junell said.
The university has about 300 server virtual machines between the three clusters. They run Nutanix’s Enterprise Cloud Platform and its Prism management software. The data center currently uses VMware’s hypervisor, but plans to switch over to the Nutanix hypervisor, AHV, in 2019.
In fact, adoption of Nutanix AHV has grown by 214 percent since the end of fiscal year 2016 to the end of fiscal year 2017 in the state, local, and education (SLED) market, the company announced today.
SLED customers choose Nutanix because it can support a wide range of workloads, according to the company.
“We run everything that can be virtualized on Nutanix,” Junell said. This includes the SQL database servers, Skype, Microsoft Office 365, the external website, unified communications, and the Banner enterprise resource planning (ERP) system. “The legacy ERP doesn’t run there because it’s very difficult to move, and the domain controllers are still in physical boxes. But everything else — you name a workload and it's virtualized.”
Making IT ‘Mildly Interesting’Using Prism to manage the Nutanix clusters makes data center infrastructure management easier because it provides a centralized view into all of the university’s systems. It also helps with capacity planning and provides visibility into dead VMs.
“Prism is a good way to do preventative maintenance,” Junell said. “You can look at what resources are being used. Anytime someone takes a VM out of a group and doesn’t clean it up, Prism tells us. It gives you meaningful information you can act on, and it really saves staff time.”
This allows the staff to focus on “more meaningful” work, he added. “In Western’s mission statement, nowhere does it say Tier 1 IT and amazing data center. What you will see is we want to teach students, we want to make their lives better.” The Nutanix HCI “gets people out of the drudgery. I’m not going to say it makes IT totally fun, but at least mildly interesting.”