Time Machines standard for backup

ANYONE who has upgraded to Apple's new version of OSX, Leopard, knows that one of its more obvious features is an invisible, automatic, built-in backup.

As long as a Mac running Leopard is periodically connected to an external drive, the operating system will capture and preserve copies of your data at different points in time.

This feature, Time Machine, has several notable aspects.

Apple's focus on a slick user experience led it to design a very simple, intuitive interface for retrieving lost files using a set-and-forget model.

Once backups are started, they just keep happening as long, as the Mac has access to an external disk.

Equally important, preserving versions of data from many points in time uses minimal space. At each backup, the software is smart enough to only make copies of files that have changed since the last backup, and to combine these with the unchanged data to present a full picture of the computer contents back to the user.

Time Machine is also smart enough to discard old versions of data once the backup drive gets full.

None of this is rocket science. It will all sound deeply familiar to any IT professional who has had anything to do with backup or network attached storage.

And Apple is certainly not alone in recognising that data on desktops is increasingly important - since people increasingly store photographs, music, documents and correspondence on them - and therefore needs to be better protected.

Microsoft has also recently improved its support for backups, via interfaces deep in Windows for third-party vendors to use when they copy the data on a disk, and its own improved backup in Vista.

But it is fair to say Windows does not yet offer anything as slick as Leopard.

But Time Machine carries a broader significance.

By building these tools into the operating system itself, and making them easy enough and good enough for home users to use, Apple is hastening the day when backup - as currently understood in enterprise IT - will largely disappear.

Time Machine is built on technologies and ideas that Apple has borrowed from enterprise IT: file system snapshots (incremental copies of changed data used to recreate depictions of the full data set at various points in time); the use of disk drives rather than tape as the backup medium; and automated policies for data capture, replication and thinning.

Since the 1990s, in fact, file storage systems from Network Appliance and other vendors have included a set of data protection capabilities broadly similar to those in Leopard, plus additional tools allowing copied data to be moved to a second machine or offsite, for redundancy.

But most enterprises do not have the luxury of starting with a clean page in IT.

Instead, they operate in a complex environment of varied applications, legacy data protection software and secondary storage devices, and established routines.

Until recently, software sold by the companies that long dominated enterprise backup - Veritas (now part of Symantec), Legato (now part of EMC), Tivoli (a division of IBM), and a few others - was clunky and outdated and did not even support using disks instead of magnetic tape for backups.

Much of the value of the software sold by these firms long rested on the availability of agents (software to copy data) for every conceivable operating system - HP/UX, Solaris, Windows 3.0, IRIX, whatever - and media managers that could drive any tape or optical library under the sun.

But as disk drives have fallen in price to the point where their raw capacity is not much more expensive than tape, and the benefit of fast writes and restores have become recognised, the game has changed.

In particular, backup to disk can be made hugely more efficient by software capable of spotting redundant data (multiple copies of the same file, for instance) and storing it only once, compared with 20 or more copies stored on tape by traditional software.

Numerous vendors now sell such systems, though none were devised by the sluggish traditional vendors.

The idea that most file systems or storage devices will eventually have backup (a systematic, robust scheme for data protection) built in as standard is still not completely accepted.

But as Apple has shown, in most computing environments there is no longer any technological barrier.

The exception is likely to remain the backup of highly transactional enterprise servers such as those used for databases, where data changes quickly and is stored as blocks not files.

Protecting this data is likely to remain the province of specialised backup software, or tools built in by application vendors.

Even traditional vendors seem to recognise enterprise backup as a separate, discrete IT process with its own costly infrastructure is a waning trend.

Instead, EMC, Symantec and IBM are increasingly focusing their backup software arms on long-term archiving, record keeping for regulatory compliance, and other more complex data management.

The disappearance of enterprise backup as traditionally conceived is still a way off, but Apple's step forward in delivering easy, automatic data protection as standard in home computing is a reminder that in the enterprise, too, times are ripe for change.

And in the opinion of most IT managers, for whom backup is still a huge pain in the neck, change cannot come soon enough.