Cranfield University has replaced its Veeam and Data Domain backup infrastructure with an infrastructure with category backup devices and Microsoft Azure cloud storage.
At the same time, the amount of on-site hardware was reduced from 24h to 4h, hardware and licensing costs were reduced and data recovery time was reduced from hours or days to minutes.
This step also gives Cranfield the certainty of disaster recovery, because Cranfield can perform all operations from any location with virtual servers on Azure when the entire site is no longer available.
This upgrade is accompanied by an upgrade in which the university replaced the existing Pure Storage flash storage arrays with 12 hyper-converged equipment nodes in the Nutanix infrastructure.
The overall project aims to simplify Cranfield’s onsite physical infrastructure during the transition, including the cloud as storage (and data processing in the event of failure).
Cranfield is a leading research institution in the field of science, industry and technology with 1600 employees and 4000 PhD students.
The computer stack is based on Microsoft and Linux servers running Microsoft and Oracle applications. In fact, it is 100% virtualized on VMware, with 400-600 virtual machines running all the time.
The existing backup infrastructure is based on Veeam backup software and data domain hardware with replication to an external data domain device.
This facility has come to an end and shows the signs, according to Edward Poll, head of the IT infrastructure.
Data Domain did what it had to do, but it was time to upgrade everything and we wanted to reduce the cost, time and complexity of management and performance improvement, he said.
The biggest problem with the data domain was the recovery. It is well absorbed, but recovery has been more problematic. It would be nice to have a restoration, but if we had to restore multiple servers – 50, 100 or 150 – we would have a hard time doing that.
Cranfield’s IT department has already started moving to the cloud with StorSimple devices – about 80TB in place and 0.5TB in the azure cloud – and has discovered how cost effective this solution can be.
The azure blue suited us well, and at first we thought we could use cases of Veeam and Data Domain in the cloud, but we were offered: Why don’t we remove the software layer, and we’ve looked at using heads, Pall said.
The category belongs to an emerging category of backup devices that take the form of nodes embedded in clusters, such as hyper-converged infrastructure.
The Software Device section can be delivered on Cisco, HPE or Dell certified server hardware with flash memory and a rotating disc inside. The capacity of at least four clusters of nodes is between 64 and 160 TB, depending on the hardware.
Customers can define a policy that determines how long data is kept as a backup and how accessible it is for the production of section material. Section backup data is treated as an NFS file share before it is sent to the internal physical archive or the cloud.
Cranfield used eight R348S section nodes with a total capacity of approximately 80 TB, containing flash drives and SAS rotary drives. The data is absorbed and then copied to the azure cloud.
The key benefits of the research team are significantly improved recovery times and the ability to ultimately restore virtual machines in the cloud, so that employees can work anywhere in the event of a disaster.
The CloudOn in the section allows for a quick recovery, enabling business continuity in the event of an accident, according to the survey. Once our site is down, we can quickly convert our archived VMs into cloud instances and if necessary run those applications in Azure, he added.
We see no difference in data collection, but the recovery performance is much better.
In terms of value, Cranfield spent £50,000 a year on off-site hosting. Today he spends about £25,000 a year on Microsoft Azure.
Meanwhile, the time needed to manage backups has been reduced from about half a day a week to five minutes a day.
In terms of saving physical space and equipment, according to a study, the university has switched off 42U storage and backup devices, of which 24U is a backup server and data domain.
Overall, this has allowed us to have a simpler, faster and more reliable backup service, he said. It is easier to integrate into the department working towards the DevOps model, and when it comes to data recovery, we are on the road in minutes instead of hours.
Storage and backup upgrades – with the transition to a hyper-converged infrastructure – are part of a broader plan to streamline IT by using modern device formats with less physical space and clouds.
The survey has been added: The university’s master plan provides for the demolition of the computer department and the closure of two large computer centres on the site. Instead, there will be a data center, a fault tolerance space for network hardware redundancy and a cloud.