When is a backup program not a backup program? A lot of software that calls itself "backup software" does not actually perform backups in the two key senses I discuss in my new ebook, "Take Control of Mac OS X Backups." That is to say, some backup programs do not create additive incremental archives of your files, some do not create bootable duplicates, and some do neither! The latter category includes, interestingly enough, Apple’s own Backup application.
Unfortunately, because software developers use terms such as "incremental," "archive," and "backup" differently, you may think you’re getting certain capabilities when you buy a product that later turn out to be missing. Thus it is extremely important that you read the fine print, and understand exactly what it is you’re looking for.
Duplication Features — Many different applications can create a bootable duplicate. This entails copying all the files (including hidden files) on your hard disk to another volume while preserving Unix ownership, permissions, and symbolic links. In most cases, such applications can also update a duplicate incrementally (rather than recopy every single file each time).
However, you should consider a few other things when looking at a duplication program:
Can it create a bootable duplicate directly onto a hard drive (as opposed to an intermediate disk image or optical media)? If you have an extra hard drive available, you’ll want this capability.
Conversely, can it create a restorable duplicate onto optical media or a disk image? Sometimes this capability is useful, other times not.
Can it automatically update the duplicates on a schedule?
When updating a duplicate incrementally, can it also delete files that were deleted on the source volume? If not, your duplicate may include extraneous files that you don’t want.
Does it have any other features you might use, such as file and folder synchronization?
That said, the duplication programs I’ve tried are more alike than different, so if you’re looking for an application to accomplish only this one task, just about any of the duplication programs I list in the ebook should do the trick.
Archiving Features — Among applications that provide archiving features, there’s a huge range of variation in how they work – and how easy they make it to restore your work later. The fact that an application stores multiple revisions of each backed-up file does not, by itself, make it good for creating archives.
Archive Varieties — First, there’s an important distinction to make: true archives versus rotating backups. In a true archive – that is, an additive incremental archive – every version of every file you designate is saved, but identical files are never duplicated. In a rotating backup, the program creates a complete, separate copy of all your files every day – basically a non-incremental archive. Then, after a certain number of days (specified by the user), the program erases the oldest backup and adds a new one. Rotating backups, because they copy every single file each day, take longer to perform and require much more storage space. If you have room and time to spare, there’s nothing wrong with that approach, and it removes the need for a snapshot list, since all the files themselves are there (see "Snapshots and File Lists," just ahead). However, because you’re erasing files older than a certain date, you’re restricting your restoration capability. If you keep, say, five days worth of rotating backups and find you need a file you deleted a week ago, you’re out of luck.
A few applications offer the best of both worlds: rotating archives. Like a conventional archive, new files are added to the backup incrementally (without overwriting older versions). However, in order to conserve space, you can opt to erase the oldest versions of selected files at the same time – for example, all versions older than 30 days, or versions copied more than 30 sessions ago.
File Format, Compression, and Encryption — To oversimplify somewhat, most software employs one of two basic methods to copy files when performing a backup. One way is to copy each file in a stand-alone Finder-readable format, so that the backed-up files look exactly like the originals. Another way is to copy all the files into a single, larger file (sometimes called an archive file or a backup set). Each approach has advantages and disadvantages.
Finder-format copies can be restored without the use of a backup program – just drag and drop. Some people also feel more secure knowing they can get at their files easily even if their backup software goes south. Of course, the backed-up files will always take up exactly as much space as the originals.
Archive files can be compressed efficiently as they’re stored, potentially saving a large amount of hard disk space. They can also be encrypted, so if your backup media is lost or stolen, no one can read your files without knowing your passphrase. And unlike Finder copies, which always take as their owner the user name of the person currently logged in, archive files can preserve original Unix ownership and permissions. Of course, you will need the backup software to restore files, and in some cases you have a slightly higher risk of data loss due to file corruption (since all the data is stored in a single file) – but most backup software has verification mechanisms to compensate for this.
(Note: Not all programs that offer compression or encryption copy data into a single archive file. A few can compress or encrypt individual files, such that they can be moved or copied (but not opened) in the Finder. You must still use the backup software to restore them to their original state.)
However, you should also be aware of a third option: disk images. Some backup software, at least when backing up to a hard disk, stores files in a disk image. Like an archive file, a disk image is a single file that contains all your other files – and can optionally be compressed, encrypted, or both. The difference is that you can double-click a disk image, and after supplying the passphrase (if necessary) it will mount on the Desktop as a regular volume – after which you can read and copy files using the Finder. Sounds great, doesn’t it? It can be, but keep in mind that in most cases, each incremental archive backup is stored on a separate disk image, so without a snapshot or file list provided by the backup software (see "Snapshots and File Lists," just ahead), restoration can be quite involved.
When making a bootable duplicate onto another hard disk, Finder copies are obviously mandatory. For archives, though, I strongly prefer a format that offers both compression and encryption – and in this respect, archive files are generally more elegant and convenient than disk images.
Snapshots and File Lists — When it comes time to restore files from an archive, you must be able to locate the versions you want quickly and easily. Some backup programs facilitate such restorations by offering snapshots – lists of all the files on your computer as they existed at the time of each backup, even if they were already present in the archive and therefore not copied during that particular session. Suppose you want to restore all the files on your machine as they existed last Tuesday. Having a list of all the files as they appeared on Tuesday – and an automated way to restore them – can be extremely valuable.
On the other hand, imagine that you want to look back at every version of just one particular file as it existed over the past month. In this case, you don’t want to wade through snapshots – you simply want a list (sorted by file name or date – or better yet, searchable) of each version of the file in the archive, from which you can choose just the ones you want. Without either a snapshot or a file list, you’ll need to locate each version of the file manually in dated folders. This makes for a long and tedious restoration process.
Sources and Destinations — The volume from which you are backing up files is known as the source; the volume to which you are backing them up is known as the destination (or target). Be sure the software you select can accommodate the sources and destinations you wish to use.
All backup programs can copy data from your startup disk. Most can also copy data from other attached hard disks, network volumes (including AppleShare volumes, FTP servers, and iDisks mounted in the Finder). And usually you can select arbitrary folders or files anywhere on those volumes to be backed up. However, there are exceptions. Qdea’s Backup Simplicity, for example, supports only your startup volume.
In most cases, your range of destination options also includes any Finder-mountable volume. (So, theoretically, you could even back up one network volume to a different network volume if you wanted to.) But not always: Babel Company’s Impression, for example, cannot copy files directly to a hard disk – though it can create a disk image that resides on a hard disk. On the other hand, at least Impression creates the disk images for you. Most programs require that you manually create the disk image yourself using Disk Utility and mount it in the Finder before you can use it as a backup destination.
A similar issue comes into play with optical media. A backup program can support recordable CDs and DVDs as a destination in either of two senses:
You pop a blank disc into your drive, give it a name, and allow it to mount in the Finder. The backup software sees the disc as a possible destination like any other volume. After running the backup program, you then return to the Finder to manually burn and eject the disc.
The backup program itself asks for blank media when needed, writing to it directly without the intervention of the Finder.
The first way of supporting optical media is trivially easy for software developers to implement, so that is how many backup programs work. But this approach does have some problems. First, it requires much more human intervention – performing manual steps despite the fact that the backup itself runs automatically on a schedule. Second, it eliminates the possibility of multisession recording (the ability to record additional chunks of information on a partially used disc after the initial write session), since the Finder does not include this feature. This is a problem because without multisession capability, you will use a much larger number of discs – increasing not only media cost, but inconvenience. (Note, however, that some applications, including Retrospect, use a packet-writing technique to add data to partially used optical discs. This is even more efficient than multisession support, but it means that only the application used to record the discs can read them later.) Therefore, if you need to record backups onto optical media, I strongly recommend using an application with multisession (or packet-writing) support.
A related issue is what I’m going to call media spanning. Suppose you have more data than will fit on a single CD or DVD – or even that you have a single file that’s too large to fit on a single disc. Some backup programs intelligently manage backups that span multiple discs, prompting you for new media when required during a backup (splitting files if necessary), and asking for the proper discs when restoring files (rejoining split files). Although the need for media spanning could affect those backing up onto hard drives as well, it’s most crucial for those using optical media. Only a few backup programs offer media spanning, and even fewer include both media spanning and multisession or packet-writing support.
Selectors and Exclusions — Selective archive backups (as opposed to full archive backups) do not include every file on your hard disk. But archiving even your entire home folder may be overkill, since it includes things like cache files, which serve no useful purpose in the context of a backup, and digital media files (such as MP3s ripped from your CD collection), which, because they change infrequently, are adequately backed up already if you maintain bootable duplicates of your entire hard disk. So instead of simply selecting one or more folders to archive, you may wish to explicitly include or exclude certain types of files.
Some backup programs include user-definable criteria specifying which files should be included (selectors) or excluded (exclusions) from a particular folder or volume – and a few programs offer both. Depending on the program, these criteria may include file names, sizes, Finder labels, extensions, modification dates, and any number of other factors.
In general, I find exclusions more useful than selectors, though I would not generally consider either an absolute must in a backup program. Your mileage, of course, may vary.
Ease of Restoration — No matter how easy it is to back up your hard disk, if your software makes it difficult to restore files, you’re going to be unhappy with it. After all, a backup that you can’t restore is worthless. Backup programs typically offer one of three main approaches to restoration:
Finder restoration: The backup program has no Restore command; to restore files, you drag them manually from the backup volume onto your hard disk. This is fine if you’re restoring an entire folder, but if you’ve done an additive incremental archive, you may have to sort through dozens or hundreds of folders to locate the right versions of each of your files.
Reverse backup: In this scheme, the backup program once again does not offer a Restore command, instead expecting that you’ll simply swap the source and destination locations and perform your backup again – in reverse. While this may reduce manual effort somewhat, it’s still going to be a hassle when restoring versioned files from an archive. And even in the best cases, a reverse backup can be confusing and stressful, because it’s easy to get the source and destination mixed up when their contents are so similar.
A Restore command: The backup program (usually) keeps track of all the files you backed up during each session, allowing you to copy them back to their proper locations – or another destination of your choice – with a few clicks. In most cases, before starting the restoration, you can choose a subset of the files, or even pick out one version of a single file if that’s all you need. Restore commands and snapshots tend to appear together.
It probably goes without saying that I prefer applications with a Restore command – they make the restoration quicker and easier. Of course, the presence of a Restore feature does not, by itself, mean the process will be easy, but it’s a hopeful sign.
Restoring a Full Archive as a Bootable Volume — If you choose to perform a full (rather than selective) archive, bear in mind that not all backup software can restore your archive from an arbitrary point to a blank disk in such a way that the resulting volume will be bootable. In order for a restored full archive to be bootable, several things must be true:
All files needed for your computer to start up – including a great many hidden files – must be included in the backup and restored afterward.
The backup software must preserve Unix ownership, permissions, and symbolic links during both the backup process and the restoration process; doing so requires that you enter an administrator’s password.
When restoring the files, the destination disk must not contain any extraneous files that could interfere with booting; normally, this implies erasing the disk before restoring the archive.
Most backup software that provides both duplication and archiving features also enables you to restore a full archive as a bootable volume, assuming that you set it up properly. Some programs, however (notably Synchronize Pro X) can restore a bootable volume only from a duplicate, not from an archive. A few applications permit full archives to be restored as bootable volumes, but lack a snapshot feature – meaning you must manually locate and copy a large number of documents to return your disk to the state you wish to recreate.
Unfortunately, most backup software does not explicitly state whether or not it can restore a full archive as a bootable volume, and of the programs that do, some are more reliable in this regard than others. This may be a good reason to consider performing selective backups instead; on the other hand, if full archives are important to you, I recommend using Retrospect.
Ease of Use — In addition to ease of restoration, an application’s overall ease of use is also important. The interface should be self-explanatory – ideally, clear enough that you can figure out how to perform a basic backup and restoration without ever looking at a manual.
If your backup software is difficult to learn or set up, you’re less likely to use it. So you want an application you can configure in an hour or so – not something that takes you an entire day to figure out. You also want your backup software to perform its duties on a schedule with as little interruption to your routine as possible. The best backup software would be completely invisible, working silently behind the scenes until you needed it.
Even so, don’t underestimate the importance of good documentation. An extensive, well-written manual can be a godsend when trying to comprehend the minutiae of rotating archives or client-server configuration.
Support and Reputation — Some backup software is published by individuals who like to program in their spare time. At the other end of the spectrum, some backup software is published by large corporations with a small army of programmers and a full-time paid technical support staff. Ironically, I’ve often received better and quicker technical support from individual authors – even those who give away their applications for free – than big companies. On the other hand, if you’re entrusting all the data on the computers in your home or small office to a backup application, you may feel more comfortable knowing that a professional staff stands behind the product.
Of special note in this regard is Dantz (now owned by EMC), developers of Retrospect. They charge $70 to speak to a technical support representative on the phone – a seemingly outrageous fee. However, I’ve used Dantz technical support more than once, and I believe you get what you pay for. The technicians answer promptly, are highly trained, and continue working with you – even over multiple phone calls – until the problem is solved (without charging you for each call). When I’m terrified that I might have just lost all my data and my software doesn’t seem to be functioning correctly, I’m only too happy to pay $70 for the reassuring voice and advice of an expert who can help me get things working again.
Price — The backup software I discuss in the ebook ranges in price from free to $130. The price does not necessarily correlate to capabilities, but I urge you not to skimp when it comes to backup software just to save a few dollars. After all, time is money. If you lose a day of income because your backup program makes you jump through too many hoops when restoring files, that’s likely to be a bigger financial hit than the cost of better software.
Take Control of Mac OS X Backups — In "Take Control of Mac OS X Backups," I take this information a step further and provide detailed recommendations about which software is best for particular uses, including network backups. I also include an appendix with feature comparisons, pricing, and contact information for about two dozen backup applications. In addition to software recommendations, I discuss hardware options, backup strategies, restoration techniques, and more – everything you need to know to set up a reliable and easy-to-use Mac OS X backup system. "Take Control of Mac OS X Backups," a 96-page ebook, costs $10; as always, purchasers are entitled to receive all minor updates free of charge.
[Joe Kissell is a San Francisco-based writer, consultant, and Mac developer who kicked off the Take Control series with the best- selling "Take Control of Upgrading to Panther," and has also written two ebooks about Apple Mail. His secret identity is Curator of Interesting Things for the Interesting Thing of the Day Web site.]