Incremental backups sound usually better than full backups because of having a mirror of your files/folders using the same disk space and depending on the number of files changed between backups can save outbound data transfer over time, but the drawbacks are many as well, specially when using a remote storage like S3, DO Spaces, etc...

At some point we experimented incremental backups as a file back up option when users created file backups where we synced their chosen directories with their chosen remote storage (I am unaware of any independent systems, but some systems like Rackspace's may use a more advanced method that keeps track of changed files and allow a restore to a particular point in time). 

The drawbacks of our incremental backups trial were:

  1. Super slow backup time when many files and folders exist - that's because of the tremendous overhead when uploading thousands of files since every file is uploaded separately in a single request. We -for sure- used to upload many files in parallel, but still each file needs a single upload process.
  2. The benefit is negated when the source directory is changing constantly.
  3. Backup restore is also time consuming since you need to download thousands of files from your remote storage.
  4. No file compression since files are uploaded as is.
  5. It did not allow a restore to an earlier time in the past. Since the remote storage always had a mirror of the source.

On the other side, we believe that full compressed backups are the best option when you need backups stored on a remote storage (these are not server snapshots, snapshots are very slow and take a LOT of unneeded space plus tie you to the cloud platform of your choice).

Some of the points we love about full compressed backups (I can tell you more of them):

  1. It's the fastest backup method. Only, your chosen files/folders are compressed on the host server and become a single file, most of the time small fraction of the whole backup size, then transferred to any remote provider without overhead. It's just a single relatively small file.
  2. To restore, you just download a single -again, relatively small file- then uncompress it, it retains the same directory structure.
  3. You can store as many copies are you want, run backups on demand to store a copy before certain events and easily switch the remote storage if you want.
Did this answer your question?