Incremental Versus Differential Backups

Even a casual home computer user can accumulate some pretty large
amounts of data that need backing up.
People may mail you large attachments that contain spreadsheets or big documents,
making your mailbox bigger and bigger over time, for example.
Even just the common hobby of taking digital photos can quickly
make your disk space requirements mount up.

Differential and incremental backups are easy ways to reduce
the time and space requirements of backups, which may encourage you
to backup more often.
This benefit increases when you want to keep multiple backups of the
same data.
For example, instead of just backing up all your files to a recordable
DVD disc every night, you could back it up onto seven different discs — one for
each day of the week.
That way, if you realized on Friday that you accidentally deleted an important
section from a document back on Tuesday,
you could just pull out the backup disc labelled
and restore that particular version of the document, even though
you’ve made other changes to the document in the days since then.

Incremental and differential backups mainly show up in file-based backup software,
not image backup software.
There are exceptions to the rule, however, and I’ll
talk about that below.

Incremental Backup

Incremental backup just means only backing up data that changed since
the last backup.
Incremental backups can be a great way to save space and yet retain multiple
versions of changed files.
I’ll give you an example to make this clear.

Remember the example I gave earlier about using seven backup discs,
one for each day of the week?
Suppose that a full backup requires 20 minutes to burn to a recordable DVD disc.
You can cut that time down significantly by using incremental backups.

With an incremental backup scheme, you might decide to make a full backup only on Sunday.
On all the other days of the week, you could just do an incremental backup, which means
only backing up the files that have changes since the previous backup.
Because probably only a small percentage of your backup data changes on any given day,
So, instead of 20 minutes to burn a backup, your Monday through Saturday backups might
just take 5 minutes each.

This incremental backup scheme can also give you a rough record of when
files changed.
Each of the incremental backup discs only contains files that changed that day,
so looking back through the discs shows you what files changed on which days.

Differential Backup

One drawback of the incremental backup comes if you need to do a full restore.
Once again, suppose you’re following the full backup on Sunday, incremental daily
backup scheme I described earlier.
But now suppose your hard drive dies on Friday; consider what you
would have to do to restore all your backed-up files.

Obviously, your most recent full backup was on Sunday, so you restore that first.
But you also have to then restore each of the most recent incremental backup discs
since Sunday: Monday, Tuesday, Wednesday, and Thursday.
That may not be too bad, but what if you were keeping a month’s worth of incremental
backups and not just a week?
Having to restore from 30 different incremental backups could be nearly as depressing
as having your hard drive die in the first place.

Enter the differential backup, which is quite similar to an incremental backup.
The difference is that an incremental backup catches files that have changes since
the last backup, but a differential backup catches all files that have changed
since the last full backup.

So now, consider a scheme in which you want to keep a month’s worth of backups.
You could perform a full backup only once, on the first of the month, and incremental
backups every other day.
But to avoid the prospect of having to restore from 30 different discs if there’s a hard drive crash,
just make one change to these scheme: perform a differential backup instead of an incremental backup
on the 7th, 14th, and 21st of the month.

What happens if you have to do a full restore on the 23rd of the month with this revised scheme?
First, you restore from the initial, full backup from the 1st of the month.
Next, you restore from the differential backup you made on the 21st of the month;
remember, that backup has everything that changed since the last full backup.
Finally, restore from the incremental backup made on the 22nd of the month.
That’s just restoring from 3 discs instead of from 22, a vast improvement.

Shopping For Incremental/Differential Backups?

Should you look for this feature when shopping for backup software?
Well, the casual home user probably doesn’t need a complex scheme of
storing different backups for every day of the month,
but since even sub-$50 backup software often provides some kind of
incremental backup, it’s worth looking for it.

Note that there is no universal agreement on terminology or functionality here,
so you have to look at a package closely to see whether it offers
incremental or differential backups, and just what form that takes.
For example, if software lets you say
"backup all files that have changed since <some date>",
then you can use that feature to create differential or incremental backups yourself (by remembering
when your last backup/incremental backup took place) even if the software has no direct support for them.

The form this feature takes may also depend on other aspects of the backup software.
For example, higher-end software may assume that you are backing up to another
hard drive or other spacious medium,
and may let you set up completely automated, complex schemes that require little
extra effort on your part (e.g., no labelling and filing recordable discs).

Incremental Image Backup

Although less common, some image backup software also offers
a feature like incremental backup.
However, since it is a disk image being backed up and not individual files,
you most commonly have only the option to restore the entire disk to
a previous backup point, not individual files.

If you take incremental image backup to the extreme,
you get what is sometimes called
"Continuous Data Protection".
With this kind of software, every little change that happens to your
hard disk is being backed up in real time to another disk
(usually onto another computer, across a network).

Another variation of this scheme involves tracking every low-level disk
change right on the same hard disk.
That gives you the ability to
"roll back" your disk to
the state it was in at any given point in the past.
This is not necessarily the best kind of backup,
but is very useful in special situations,
such as testing new software; after installing and testing a software package,
you can rollback the disk to its original state
so you can test again.