CrashPlan and Time Machine complement each other very well and can be used on the same computer, side by side, backing up your live data. But you don't want to try to integrate the two. Read on for the nitty-gritty technical details.

The short answer is that backing up Time Machine data with CrashPlan does not work very well. There is no advantage to sending your Time Machine data offsite if you're using already using CrashPlan to back up remotely.

The long, technical answer that follows describes results when we tested backing up Time Machine data with CrashPlan. This chart compares the Time Machine backup archive size before and after backing up to CrashPlan.

Because it's not practical to make a full copy of the file system every hour, Time Machine works by creating hard links to directories and data that have not changed since the previous backup.

(Skip this section if you already understand how hard links work.)

Essentially, a hard link is a file name and a pointer to the actual data. When a file no longer has any pointers (for example, when a file is deleted), the operating system allocates the space for re-use. Beginning with MacOSX 10.5 (Leopard), the OS allows hard links to directories, which is unique to OSX and is a capability Time Machine exploits.

CrashPlan follows hard links. This means that if a file has 10 hard links, the file count increases by 10 and the file selection size increase ten fold. When CrashPlan follows those 10 hard links, the file count is multiplied by 10 and the file size of every file in the directory also is multiplied by 10. That's a lot for CrashPlan to scan! After running Time Machine for just one week, CrashPlan said the Time Machine volume had 7,984,818 files (includes both files and hard links) and had grown to 303.5 GB! Mac OS measured my Time Machine volume at 63 GB (hard links excluded).

What matters is not the discrepancy between Time Machine and the OS; rather in following those hard links, CrashPlan is scanning an abnormally large number of files rather frequently. Remember, there were almost 8 million files after only a week of using Time Machine. Depending on your CrashPlan settings, this scanning could take a very long time or seriously bog down your system.

As this test illustrates, Time Machine and CrashPlan can work very well side-by-side. Just don't try to integrate the two.

Time Machine was designed for local backup only, while CrashPlan was designed for local and remote backup. If you want remote backup and just one system, you'll probably be happiest using CrashPlan both onsite and offsite. If you prefer to use two systems, use CrashPlan for remote backup of your files only (not Time Machine data) and use Time Machine (or Time Machine and CrashPlan) for local backup.

To make sure you're not including Time Machine data in your CrashPlan backups, look at your Settings screen, and see if the box is checked next to your Time Machine destination. It should look like this:

When you specify a volume as a Time Machine destination it erases the volume first. So in order for CrashPlan/Time Machine to work on the same volume set up Time Machine first and then point CrashPlan to the volume as a backup destination.

If you're interested in learning more about how Time Machine works, James Pond wrote a great article explaining hard links:

http://pondini.org/TM/Works.html