Time to take a fresh look at backup strategies
November 8, 2013
One of the onerous chores the IT department of any organization faces is backup, yet it’s a vital task for survival.
Without safely preserved copies of data, if (when) there is system disruption most businesses face extinction without information from accounting, email, human resources, supply chain … well, the list goes on.
Yet some organizations leave backup and recovery (and its brother, disaster recovery) as less than important duties to be modernized.
David Cohen found that out when he took over as vice-president of information technology over a year ago at Toronto-based CAPREIT, the Canadian Apartment Properties Real Estate Investment Trust, which calls itself one of the country’s biggest landlord.
It oversees some 38,000 apartments, with a lot of legal documents in its data centre.
Cohen figures his data is growing at about 26 per cent a year.
Yet the tape-based backup and recovery system required staff staying up on Saturday nights sometimes until 4 a.m. “And if I ever had to exercise it I would have been very, very worried about my return to operations. It was measured in days, if everything went perfectly.”
David Furst, senior vice-president of technology at Canada NewsWire, which distributes thousands of press releases a day, found himself in a similar position. The company’s EMC-based Networker solution was taking too long to back up virtual machines, and recovery time was increasingly insufficient for time-sensitive customers like those in the financial and tech sectors.
Both chose solutions from Massachusetts-based startup Actifio, which is an illustration of how the normally staid backup and recovery market is being shaken up.
The big names in enterprise backup are ASG, CommVault, EMC, IBM, Hewlett-Packard and Symantec, according to Rachael Dines, an infrastructure and operations analyst at Forrester Research. But new thinking on approaches to backup and cloud computing are disrupting the market.
An increasing number of organizations are storing some data in the cloud, she said. Meanwhile backup solution providers are adding disaster recovery features to their suites, such as data continuity and resiliency and even the ability to boot virtual machines from backups for rapid recovery.
But dealing with the astonishing rise in data growth is also a factor. The storage networks of organizations are stuffed with copies of data made by business units and branches that also contributes to lengthy backup and recovery times.
IDC estimates unnecessary storage of multiple copies of data will cost businesses around the world about $44 billion this year.
Actifio has attracted attention for its strategy of essentially having organizations keep one set of what it calls production data, and then backing up only information that changes.
The copied data includes metadata on what application wrote the data, when the backup was created, if there is a service level agreement attached to data. This makes it faster for recovery, says the company.
But the solution also includes the ability to use backed up data for quickly provisioning test and development environments for creating new applications. That’s one feature that appealed to Cohen and Furst.
Dines said new thinking like this among startups is causing other backup and recovery companies to take notice. Hitachi Data Systems’ new HDIM (Data Instance Manager) seems to be one response to the idea of not backing up all the copies of data in storage, she said.
When an enterprise is searching for a backup solution it should consider the ability to support critical applications and software platforms – including hypervisors — Dines said.
“Deep application integration and awareness is still an evolving area for backup software,” she said, “and it’s really important that your solution can adequately protect and be able to quickly recover critical applications.”
The ability to integration with the organization’s server and storage hardware, including disk libraries, is also vital, she said, as is the ability to control and catalogue snapshots and replication need to be looked at.