stefanovi - Fotolia
Black Country-based specialist steel supplier Carrs Tool Steels has dumped its legacy server and direct-attached storage IT infrastructure for a Scale Computing hyper-converged cluster.
The move allowed it to sidestep a potentially disastrous lack of resiliency without needing to spend large amounts of money on complex SAN storage.
It chose Scale’s HC1150 nodes over hyper-converged products from Nutanix because they suited the scale of the organisation better and were less costly.
Carrs Tool Steels, based in Tipton, West Midlands, supplies tool steels to about 1,000 customers ranging from micro-SMEs to car manufacturers, including Rolls-Royce. The company specialises in providing pre-machined tool steel blocks.
All processes in its factory are controlled and monitored via its Priority enterprise resource planning (ERP) system, which was running on a physical server infrastructure with direct-attached storage.
There was no redundancy across this configuration and the company had suffered outages that meant its shopfloor was completely unable to function, said IT manager Mark Mullaly.
“Lack of redundancy was a major issue, as was performance,” he said. “It was old and full, and just messy. It was difficult to control and to get an idea of how things were running.”
One outage stopped production for most of a working day and this highlighted to the board the need to upgrade.
“It was clear that a simple thing could bring down the entire IT infrastructure,” said Mullaly.
At first, he looked into replacing the servers with newer ones, but it soon became apparent that a SAN would also be needed, which would push things to new levels of cost and complexity.
“I’m not a hardware specialist,” said Mullaly. “I didn’t want to get too involved in the technicalities of networks and storage products and spend three weeks a year on training courses.”
Read more on hyper-converged
- The pros and cons of file and block access storage arrays vs hyper-converged all-in-one server/storage architecture for storage of virtual machine systems and data.
- The most-read hyper-converged tips of the past year represent a range of best practices for procuring and implementing HCI as all or part of your overall datacentre architecture.
So he started to look at hyper-converged infrastructure, which is particularly attractive to SMEs because it provides compute and storage in nodes that can be built into grid-like clusters with a virtualisation hypervisor built in.
HPE’s Simplivity and Nutanix’s HCI products were evaluated, but both came out as very expensive.
When Scale’s hyper-converged products were evaluated, Mullaly thought they were best attuned to the needs of Carrs’ business.
The company has deployed three HC1150D nodes with a total of 21TB of raw storage, which translates to about 10TB usable. Some 1TB of flash is deployed in each node, with spinning disk HDDs making up the rest of the capacity.
Backups are sent off site via a Barracuda backup appliance.
The big benefits for Mullaly are Scale hyper-converged’s resilience and ease of use. “I can spin up new servers easily and always know things are running as efficiently as possible on half a day’s training,” he said. “The system balances itself. It just works.”