The University of Reading’s IT team plans to “pull the plug” on its Nutanix-based infrastructure to test disaster recovery readiness after a move to the supplier’s NC2 cloud services.
That’s according to head of operations Kevin Mortimer, who spoke to Computer Weekly at Nutanix’s .Next 2025 event in the US last week.
The University of Reading has been a Nutanix customer since 2017. It has more than 50 research centres in areas that include agricultural, biological and physical sciences, as well as meteorology. The university has around 5,000 faculty members and 20,000 students.
It runs Nutanix Enterprise Cloud Software on Dell EMC XC Series hyper-converged systems, with the Nutanix AHV hypervisor.
The plan now is to use Nutanix’s NC2 cloud provision as failover for on-site activity.
When it’s up and running, Mortimer plans to pull a plug to test it – literally.
The university already uses Rubrik for backup, but it wanted a way to fail over in a disaster recovery (DR) scenario. The question for Mortimer was: “If we lost our main datacentre on campus, then what do we do? We have a few IaaS [infrastructure as a service] workloads in Azure, and things like domain controllers that are really critical.”
But to put all those into Azure was going to cost up to five or six times more than it would if it was on-premise, according to Mortimer.
“So if on-premise was £100,000, public cloud would be £1m,” he said. “But NC2 is kind of in the middle of that, so you’re getting the benefit of some cloud and it does cost you more, because cloud is generally more expensive.”
Nutanix Cloud Clusters (NC2) is a hybrid cloud platform that provides a consistent user interface to all environments.
“It’s a step up from colo, but you get the benefit of public cloud networking and the management piece,” said Mortimer.
Disaster recovery test
Mortimer’s team is still working through the details of NC2 deployment, but one benefit he’s looking forward to is to be able to do a full DR test.
“That’s something that is always quite scary to an organisation, but it’s something I’m really keen to do,” he said. “We can turn it off. We know the processes. We’ll let it run from the cloud for a week beforehand so it’s not all last-minute, and we can fail it back.
“I’ve threatened my team,” said Mortimer. “I said … ‘I’m going to walk into the datacentre, and get a cable and pull it’. Because in a real-life scenario, that’s what could happen. If it doesn’t work, at least you can know it isn’t working.”
That’s likely to happen this summer or next Christmas, to avoid periods of heavy user activity.
According to Mortimer, it will involve about 60TB of data that will be closely synchronised in terms of RPO.
Key steps
What are the key steps for Mortimer in the run up to that kind of test?
“Once the platform’s all stood up and connected, we need to make sure we can move workloads between the two clusters,” he said. “That’s validating the networking, which is going to be the trickiest part. It’s things like making sure DNS is available, and tier-zero services like Active Directory.
“But once we’re confident that networking works for those test cases, and if everything’s been rebuilt out from scratch against those principles, then it should be just a case of pull the plug and fail over,” said Mortimer.
The university migrated from VMware to AHV when it moved to Nutanix, and later deployed Enterprise Cloud to build a self-service portal to allow academics to provision their own resources where they can manage virtual machines (VMs) and storage directly, similarly to a public cloud platform.
In total, it now has around 600 VMs that run across eight nodes in its datacentre, and something like 60 applications.
It also uses Nutanix Files, manages things with the Nutanix Prism user interface, and has added Self-Service portal and the Nutanix Calm management framework for hybrid clouds.
