"We needed to give some structure and standardisation to the storage, and the IBM SAN reaching its end of life gave [us] the opportunity to do that," said Turlough Farrelly, IT manager at Lakeland Dairies. "Management of storage was difficult, and we needed to get it under control and to introduce some simplicity to the whole process."
He added, "Storage and tape management was dispersed, and there was an overhead on operations management, with different backup regimes and tape infrastructures that were all costing time and money."
Under the firm's backup arrangements, recovering from a total loss of data using tape and disaster recovery services would take five days, with a recovery point objective (RPO) of 12 hours to 24 hours prior to failure.
In the technology refresh project, Lakeland Dairies reduced 35 physical servers to just two servers using Microsoft Hyper-V as a virtualization platform. Storage to support the virtualized servers is two Compellent Storage Center SANs at the main and disaster recovery sites in Cavan. For tiered storage, disk is split evenly between Fibre Channel (FC) and SATA at the primary site, and is all SATA at the disaster recovery location.
The Compellent SAN now acts as shared storage for all Lakeland Dairies offices, with connectivity supplied by a WAN upgrade that provides connections of between 2 Mbps (via Symmetric DSL) and 100 Mbps (via microwave) between sites.
Compellent SAN wins on tiered storage and ability to expand capacity
"We have been an IBM house and still use them for servers," Farrelly said. "We looked at IBM and NetApp storage, and also briefly at EMC, but we had no experience of them so they dropped off the short list early. With a lot of storage products you can only expand capacity so far before you need to rip and replace the entire chassis. With Compellent, you can keep adding disk capacity."
Prior to the technology refresh, data was backed up to tape that was then taken to a remote location. "To restore in case of disaster, we'd have to recover data from the tapes onto new servers, a process which would have taken about five days," Farrelly said. "Now we replicate overnight, with snapshots going between the Compellent SANs. We can now recover our Wintel footprint in minutes instead of the five days it would have taken before."
Currently, data is being backed up from the secondary SAN to tape using Microsoft Data Protection Manager (DPM). Farrelly said the firm will evaluate disk-based backup to a virtual tape library (VTL) over the next 12 months.
Energy savings from fewer spinning disks
The project has allowed Lakeland Dairies to save approximately €43,000 in total. Of this, €30,000 is money the company doesn't have to give to a disaster recovery services provider now that disaster recovery has been taken in-house. The other €13,000 comes from energy savings due to the smaller number of spinning disks and processor load owing to virtualized servers and shared storage.
"We're based in the countryside, and we have to think about our green impact, so we looked at what we might be able to save," Farrelly said. "With the whole technology refresh, we now take up 80% less rack space, we save 38 tonnes of CO2 a year, and are making an annual saving of 88% in power and cooling use."