The backup challenge

BitDepth#1495

Mark Lyndersay

IT'S BEEN a while since backing up your data was a topic in this space.

Six weeks ago, I tried to recover files from a camera memory card. After deep-scanning the solid-state memory, many files were found, but not the ones the owner was looking for.

Deletion marks data as erased, but doesn't actually remove it. It's a bit like removing the label from an intercom list in an apartment complex. The data, like the occupant, is still there, but the computer has no way to know that, so it will write data into the space. Eventually, new data destroys the old. That's why Department of Defense (DoD)-level formatting takes so long. Random data is written repeatedly into drive space marked as empty to destroy what's there.

Four weeks ago, I deployed a specialist software tool to recover data from a Parallel ATA (PATA) drive from a 15-year-old computer system.

Three weeks ago, I watched a key workstation hard drive disappear from my computer desktop.

That failure was mortal and complete, taking with it three years of photography assignments in an instant.

It took four hours to restore from a continuous backup system, but I did lose the file I was working on when the hardware failed.

Reviewing the status of other drives in my system, it became clear that two others were approaching end-of-life and I needed to move the data on them to new storage.

At some point, it became clear that the universe was done whispering in my ear about reminding readers about backup and began to holler.

Here are some things you should know about lost data.

All the money you think you're saving by not buying additional hard drives will be lost ten times over paying professionals to recover data from even a single digital unit.

An effective backup system is insurance bought to protect against an inevitable failure that will cost you.

Yet far too many people think of backup, its cost and fussy implementation as expensive, possibly overkill and too much work to invest against the "possibility" of catastrophic loss.

So here's what I've learned through pain and crisis.

Digital storage, spinning hard drives, solid state drives (SSD), sophisticated tape systems and optical media all have a useful life.

From the minute we create and save digital files, a timer starts that's invisible to most.

Modern hard drives and SSDs are mechanisms managed by integrated circuits and part of their programming includes SMART (Self-Monitoring, Analysis, and Reporting Technology) monitoring, which tracks several hardware parameters and indicators.

If you have oversight responsibility for multiple drives or just a specific paranoia about a single unit, investing in software that accurately reads SMART status gives you an important window into the deterioration of your storage.

Without that insight, you are left to consider more terrifying warnings like copy operations that abort, drives that spontaneously dismount, sluggish drive access and the worst of all possible alerts, the clatter or scraping of a drive head making contact with the polished platter's surface.

The most basic architecture of an effective backup system is called the 3-2-1 backup rule.

A proper implementation of this requires that you have three distinct copies of your data. Two of those backups should be on different kinds of media (an external drive and cloud storage, for instance) and one backup should be held offsite.

Time is not your friend. A hard drive that's left unused on a shelf risks mechanical seizure; one that's in continuous use will face running wear as the hours of use increase.

Most drives that aren't accessed in a computer system will park their heads at rest, but they are powered on and reporting their SMART status.

It would be risky to trust a spinning hard drive that's been in use for more than five years.

By then, capacities will also have increased and the cost per gigabyte will have dropped, so a data migration budget can buy larger disks, making it possible to consolidate data.

But those primary disks must be set up to synchronise their data with other storage units in the backup system. A backup, by definition, exists in at least two places.

Automate these processes as far as possible. It's better to set up software to run backup processes and synchronise data across multiple instances than to pretend that you have either the time or inclination to do it yourself.

You probably won't do what's required, you certainly won't do it consistently, and you probably won't realise what you've missed until it's too late.

There are only two kinds of computer users in the world. Those who have lost data and those who will.

Mark Lyndersay is the editor of technewstt.com. An expanded version of this column can be found there