Sixth form college rebuilds after fire with Scale hyper-converged

A Manchester sixth form college has ditched VMware and shared storage for Scale Computing hyper-converged infrastructure with the company’s KVM-based hypervisor.

The move came at Xaverian College – which has 2,400 students, 250 staff and 1,100 PCs/endpoints – after a fire wiped out its existing storage infrastructure.

The college virtualised all its applications in 2010, including student administration and data, Office apps and specialised software for media, IT and music students.

At that time, the college opted for VMware virtualisation and this necessitated shared storage, which was provided by Scale Computing, in its early incarnations, with multiprotocol storage (but not hyper-converged).

Then, in 2014, the college suffered a server room fire, which precipitated a move to hyper-converged infrastructure, said Martyn Bratt, networking and IT resources manager at Xaverian College.

“It wasn’t the fire that damaged the kit,” he said. “It was the loss of power. This resulted in a loss of drives, at least one in each node, with the same in the disaster recovery cluster, so the hardware became unsupported.”

By that time, Scale had moved on from selling storage-only hardware, and “was the obvious choice to move to hyper-converged infrastructure”, said Bratt.

“The move to HC3 was a no-brainer because of its simplicity of use and licence costs compared to VMware, which is great for larger organisations, but no so much for our size of business,” he added.

Bratt’s team migrated the college’s entire VMware server estate to Scale KVM using DoubleTake. This totalled about 35 virtual machines on three physical servers. These are mirrored to three more that serve as a disaster recovery location, on the same site but physically remote.

HC3 offers built-in disaster recovery using high availability, cloning, replication and snapshots.

Meanwhile, the college’s IT team used DoubleTake to repurpose the old Scale hardware into an unsupported “dirty storage” cluster for its sandbox environment with Hyper-V, “because it’s free”, said Bratt.

The Scale Computing deployment at Xaverian now comprises five nodes of hyper-converged nodes at primary and DR locations, with 28TB of usable capacity in each cluster. The most recent addition saw a four-node cluster of HC1150s deployed to production.

Hyper-converged infrastructure is a rapidly emerging category of product that combines compute and storage in a single node, but builds these into clusters – with built-in virtualisation hypervisor – and grid-like scaling across and between nodes.

For these reasons, hyper-converged is easy to deploy and well-suited to smaller organisations that may lack large and specialised IT staffs.

Bratt said: “Hyper-converged seemed an obvious way to go with less hardware, and less demand on power, cooling and space.

“We can’t easily quantify the benefits with figures, but for our team of five, it means we can spend more time on first-line support, which is most of what we have to do.”