Gordon Laing

Like all types of insurance, backing up your data can feel like an inconvenience until the day you need it. I survived my first disk failure by the skin of my teeth: long ago I entrusted my digital photo collection to a SCSI hard disk and backed up occasionally. One day the disk failed to start and, as I tried to remember exactly when the last backup took place, I had a terrible sinking feeling. Just how many photos had I lost forever?

Ever the optimist, or perhaps in sheer denial, I reconnected the disk from time to time to see if it would magically start up again. The first few times it either clicked ominously or did nothing at all; then one day, amazingly, it powered up just long enough for me to safely copy the contents with seconds to spare. I couldn’t believe my luck, and from that point on took it as a message from the backup gods to give my data the regular protection it deserved.

It only takes one loss or a lucky escape to realise that even if you do back up, you are only protected to the point this was last performed. That could be last night, last week or even last month. Whichever, any new data produced between that time and the point when disaster strikes is potentially lost.

Once you realise this you may decide even daily backups are insufficient. What we need is a form of constant backup, not just daily or even hourly, but one happening all the time. You would also want that backup to take over the instant a problem occurs, allowing you to continue without downtime.

This sounds like a tall order, but this is exactly what Raid arrays deliver. They employ redundancy, most commonly Raid level 1, to constantly duplicate data from one disk to another so, should one disk fail, the other instantly takes over, avoiding downtime while the broken hardware is replaced.

I’ve written many articles about Raid, benchmarked various configurations, simulated disk failures to monitor arrays rebuilding, and used it on my systems for years, but until recently I’d never experienced its protection for real. Then last month I had two disk failures within days of each other and both involved what I’d describe as critical data, but thanks to Raid nothing was lost.

The first failure involved my main office PC that uses a three-drive, hardware-accelerated Raid 5 array to house my photos, documents and emails. The second involved my dedicated off-site server that uses a twin-drive Raid 1 array to host a web business. Both systems are backed up daily, but any downtime on either could involve losing personal items, work and revenue.

When a disk failed in my office system, the Raid management software warned that the array was in a critical condition and would remain so until the broken hardware was replaced. Luckily I had a spare disk, so I swapped it and allowed the array to rebuild.

An ocean away, the host service managing my server alerted me of the disk failure shortly after it occurred, along with the reassuring news that it had already replaced the broken disk as per our contract.

In both cases I was struck by the total lack of stress and hassle ­ there was no data loss, no downtime, and I would not have known about the server problem if I hadn’t looked at the logs or the host hadn’t informed me.

Before implementing Raid, my disk failures evoked panic as I checked the contents of the most recent backup, followed by downtime as the data was restored. In contrast, these latest disk failures passed with barely any impact.

It is important to remember that Raid can’t protect against fire, flood or theft, or files deleted by mistake, so regular backups (preferably kept elsewhere) remain an essential part of protecting your data. The discretion with which Raid operates in the background makes you wonder if a regular backup would be sufficient by itself ­ the thought crossed my mind a few times during years of Raid operation without incident.

But after experiencing the real-time protection and zero downtime of Raid, I’m convinced that anyone who takes their data seriously should include it as part of their backup strategy.

Like this story? Spread the news by clicking below:

 del.icio.us     Digg this     reddit!