ake1150 - Fotolia
The University of Reading has consolidated numerous legacy storage systems onto Nutanix hyper-converged infrastructure, with around 600TB of storage capacity and cloud-style self-service operations for its in-house customers.
In the process, it has cut hardware expansion and virtual machine (VM) and storage provisioning time from weeks to minutes and rack space from five to one, with consequent power savings.
The Nutanix cluster – which cost between £300,000 and £400,000 – replaces an existing infrastructure made up of numerous storage arrays and based around a Red Hat-supplied GlusterFS file system (covered by ComputerWeekly in 2013) that had become cumbersome and difficult to manage.
The Nutanix cluster that forms the Reading Research Cloud was built to support meteorological and medical data and imaging with very random input/output (I/O) requirements, and has generated about 0.5pb of data so far.
“It was built with the aim of giving the most bangs for the buck but we started to suffer a massive failure rate,” said academic computing team manager Ryan Kennedy. GlusterFS was not designed for the workload it eventually had to handle.
“It would take weeks to scale to add more storage. It was mostly about how it was designed when we implemented it. It wasn’t designed to scale to petabytes and it couldn’t cope – and we couldn’t retrofit it to Red Hat best practice guidelines.”
Kennedy’s team has deployed Nutanix hyper-converged infrastructure on Dell EMC XC appliances running the Nutanix AHV hypervisor. It provides around 50 compute virtual machines and 60 storage instances via VMs.
More on hyper-converged
- The rise of hyper-converged infrastructure – with compute, storage and networks in one box – seems ideal for SMEs, but is it always a better idea than traditional IT architecture?
- Converged and hyper-converged infrastructure are ready-tested ways to put compute and storage into the datacentre. But which should you choose for your deployment?
It is now the department’s only storage system, and its easy scaling is a great benefit because the university has only a small IT team and storage array knowledge comes at a cost.
The university opted to use the Nutanix AHV hypervisor and saved around £500,00 by doing so rather than using VMware. “AHV comes with the hardware and we haven’t found anything it can’t handle,” said Kennedy.
The Nutanix Self Service Portal allows academics to set up the required central processing unit (CPU) and storage resources for their projects. They can configure what they need in two or three minutes where it would previously take two or three weeks.
Key benefits for Kennedy are ease of use and the ability to manage everything from a single pane of glass. “We can grow quite quickly,” he said. “Tomorrow, an academic could need the equivalent of a couple of hardware nodes. Now, we can add that easily, whereas before it would have taken weeks – and then we’d still be keeping our fingers crossed.”
Meanwhile, physical space used has been reduced from five full racks to the equivalent of just one, with consequent reductions in power usage.