The Chartered Institute of Personnel and Development (CIPD) has ditched tape backup – which uses more than 20 tapes per day – in favour of almost 100TB of ExaGrid disk-based backup data-deduplication hardware. The move has cut data restore times from hours or days to seconds.
The CIPD serves 160,000 members from four datacentres, one at its HQ in Wimbledon, plus secondary sites at Portsmouth, Heathrow and north London. With 22 IT staff, it serves 400 users with in-house-developed membership and management apps and Microsoft Exchange email. It runs these on 74 physical servers plus 20 VMware hosts, totalling around 450 servers in all.
Previously, the CIPD had an entirely tape-based backup infrastructure, with a mix of LTO standards and backup software – Symantec Backup Exec and NetBackup – on HP MSL tape libraries. In this environment it made full backups every day with a backup window of six to eight hours for 32TB of data that filled more than 20 tapes nightly.
Mark Boardman, infrastructure manager with the CIPD, said: “Tape is a cheap media. Backup wasn’t slow and the window wasn’t massive but restore times were horrendous. If we went to a blank disaster recovery site it would take hours to catalogue the tapes.”
Last year the CIPD decided to move on from its incumbent disaster recovery provider and at the same time to move to disk-based backup. It looked at a number of disk-to-disk backup product providers before settling on Exagrid, said Boardman.
More on data deduplication
- Definition: Data deduplication
- Definition: Global data deduplication
- Taking a closer look at deduplicating backup systems
- Speeding up backup deduplication
- Using deduplication when backing up virtual machines
- Is this the end for target deduplication storage appliances?
“We looked at Data Domain but everyone nearly fell off their chair when they heard the price. There was also Symantec’s NetBackup-in-a-box appliance, which we tested but found to be too much of a black box with a lack of visibility of what was going on. Also, we weren’t convinced about Symantec as a hardware vendor at the time.”
CIPD deployed three Exagrid EX13000Es, two at the Wimbledon main site and one at a DR site. Each one at the main site held 13TB useable capacity while the third was configured to 26TB by configuring the pre-processing “landing zone” to be smaller, a benefit afforded by the fact data had been duplicated at the main site prior to copying over.
Exagrid is post-process data deduplication hardware in which un-reduced data is ingested prior to data reduction taking place. Data deduplication removes duplicate chunks of data and replaces them with pointers to the original.
Customers usually find data deduplication ratios increase as time goes by as there is a greater set of data to deduplicate against. The CIPD averages around 9:1 currently, said Boardman, but has seen ratios of up to 25:1.
Since the initial deployment the CIPD has added three more EX13000Es at its main datacentre and one at the DR site.
Boardman reports benefits that include doing away with onerous tape management tasks, a lack of worries about tapes in transit, no more broken tapes and that restore is much quicker.
“There’s no setting up of tapes needed, no indexing. We get data back over the LAN at wirespeed, just the same as it went in.”