March 10, 2009, 3:35 PM — Enterprises don't like to discuss how often backups fail to take place, restorations prove unsuccessful and damaged media ships out for data recovery.
"If you discover data loss, you don't necessarily want to broadcast that," said John Sloan, senior research analyst at Info-Tech Research Group.
Data recovery firms, however, aren't pressed for business.
On average, Kroll Ontrack Inc. performs 50,000 recoveries each year across 24 countries, said Canadian operations manager John Riddell. This number encompasses everything from end users to enterprise servers, he pointed out.
Markham, ON-based CBL Data Recovery Technologies Inc., which has 17 labs worldwide, completed 21,000 projects last year, according to president and CEO William Margeson.
Enterprise recovery projects usually result from the physical failure of hard drives coming from RAID setups, noted Riddell.
"If they have one drive fail, they're okay ... but generally, when we're seeing more than two drives failing at a time, they're calling us and saying they can't restore their RAID because they have lost multiple drives," said Riddell.
And there isn't much enterprise can do to prevent this.
"There's no real cause that we can determine," Riddell continued. "Sometimes it's heat, sometimes it's the age of the hard drives, but there's no way of preventing the drive from failing. That's why there is a redundancy in place to try and prevent it as much as possible."
But the following advice from data recovery experts and analysts will help enterprises from making the problems worse.
1) Verify your backups and focus on restoration
Enterprises must confirm their backups are taking place, suggested Margeson, who recommended performing "backup fire drills."
"If you want to find out how protected you really are, walk in one day unannounced and demand to see a restoration of some backed-up data. All of a sudden, the realities will be flushed out," said Margeson.
Responses will include, "Bob, our IT guy, is not here today, we can't do it," "We don't know where that tape is," and "Wait a minute, we don't know how to restore this old backup technology because we've upgraded," he said.
"Verification of backups is critical for companies," echoed Riddell. "We've had calls from clients where they are doing everything they're supposed to. They go to restore a backup and what they're not doing is verifying their backups to ensure that what they're actually trying to backup is there and ready to restore should they need it."
But making successful backups is meaningless unless enterprises can prove they can get the data back -- and get it back in a timely fashion, Sloan pointed out. "Just saying, 'We can get it back,' often isn't enough," he said. "It has to be back in a certain time frame or you're going to start losing money."
Focusing exclusively on the backup is a mistake, warned Sloan. "When you focus on the backup you say, 'We have to back up X number of terabytes once a week' and your measure of success is whether the backup ran. The real issue is, 'If we need to get that back, how long will it take and how likely is it that it's going to be there?'" he said.
2) Backup data from desktops AND servers
"The vast majority of organizations are backing up, but they are primarily backing up on servers," noted James Quin, senior research analyst at Info-Tech Research Group. "They are not doing a tremendous amount of backup on desktops."
This is a problem because while corporations may believe there is nothing really important on the desktop, studies have shown that a lot of data isn't in structured format, said Quin.
Roughly 50 per cent of corporate data in North America is structured and 50 per cent is unstructured -- and these are conservative metrics, Quin pointed out. Of the 50 per cent that's unstructured, half is in e-mail and the other half is on things like workstations and files shares, he said.
"Some organizations do take backups of their desktops, but it's a very small percentage," said Quin. "This is probably around 20 to 25 per cent."
3) Consider restore speed and reliability when purchasing new backup media
A lot of enterprises looking at more efficient ways of performing backups, such as disk-to-disk solutions and virtual tape libraries, are doing so because they find their media isn't as reliable as they hoped or it's taking too long to perform backups to tape, according to Sloan.
Enterprises determine how much data they want to back up every night and find they aren't able to backup everything in the amount of time they are given, Sloan explained. "Taking too long to do backup to tape ... might be a trigger for looking at those technologies, but the value of those technologies is in the restore side," he said.
When enterprises explore alternative backup technologies, they should ask whether it will help perform the restores faster and whether the reliability of those restores will be better from a disk media, Sloan suggested.
4) Determine what data you can afford to lose
Changes in U.S. court procedures for civil suits over the last few years has driven a huge surge of interest in the efficiency of backup software, media and capacity, said Sloan.
"People realize they have to keep important documents and data for a certain retention period and they can't just blame the technology if they can't find it. So that's really been driving a lot of investment and strategizing around archiving and backup," Sloan explained.
But saving every bit of data and keeping it indefinitely isn't the answer. "Your objective is to make sure that you don't lose something that's really important, but you also can't meet that objective by just saving everything," said Sloan.
According to Sloan, a good backup and archiving strategy will answer the following questions: What is important? What needs to be kept? What can we afford not to have?
"The data loss prevention issue is more than just saying, 'We have to keep everything forever and make sure we never lose anything,'" said Sloan. "It's saying, 'Of all the stuff we have, what can we afford to lose in a disaster situation? What can we afford to delete for archiving purposes?' and then making sure the stuff that's left over is appropriately protected and backed up and archived."
5) Add a data recovery component to your disaster recovery plan
Including a data recovery component to your disaster recovery plan will help prevent making hasty decisions "under the gun," Riddell suggested.
"Most companies now make sure that they know what they have to do to get back and running in as little time as possible so they don't lose any money," said Riddell. "We want to ensure they have a data recovery component within that so should something happen, they can resort to their plan and know they can call a reputable company before they need them."
6) Know when to "throw in the towel"
IT staff often make the mistake of not knowing when to give up, said Margeson. "They don't know when to throw in the towel," he said. "They make the disaster bigger, with good intentions."
Enterprises shouldn't expect their IT staff to be data recovery experts either, Margeson added. "How do you know how to fix something if you don't really know what's wrong? You can't expect your IT staff to know everything," he said.
One obvious indicator of a hard drive crash is sound, said Riddell, such as a clicking, scratching or squealing coming from the server. "That's when you want to shut down right away because if you keep going, what you're doing is destroying data," he explained. "The longer you try and work moving physical drives, the greater the likelihood that you're going to lose more data."
7) Find a reputable service -- how to distinguish the good from the bad
A reputable firm will provide an up-front quote, according to Riddell, rather than ask you to send in the media first. "We already know from experience, in all likelihood, what we are going to have to do to recover that data," he said.
The estimate should also indicate the data you will get back for that price, added Riddell. "Ontrack will supply a full comprehensive file listing to the customer after we evaluate the system to make sure they know what data they are going to get back when we quote the price, before they make any decisions," he said.
There shouldn't be any hidden costs, said Margeson. "Data recovery can be expensive -- that's what has drawn a lot people into thinking and pretending they can recover data -- but you need to move with assurance ... don't pay for diagnostics, don't pay for parts ... if they are a real data recovery company, they have the inventory at their disposal," he said.
If you call a data recovery firm and they put you off or have to call you back, that's a bad sign, warned Riddell. "A good data service rep will have the answers right away," he said. "If they don't, they'll get someone on the phone with you that knows the answers. You want to make sure that they are able to answer your questions right off the bat."
About 75 per cent of CBL's projects last year involved defective disk drives, Margeson pointed out. "There isn't any software in the world that can repair or get data from a defective drive," he said.
Finding a good data recovery firm starts with referrals from friends and colleagues, said Margeson, but it's important to find a company that can cope with "just about anything."