3 ms·

Amazon S3. cron job -> tar gz -> s3cmd -> s3 Their cheap, very reliable, and fast. What else can I ask for.

• cperciva 17y ago
  What else can I ask for? How about faster, cheaper, and more flexible? If you back up the same or overlapping sets of files, tarsnap's snapshotting functionality will probably result in it using less bandwidth (and thus end up being cheaper) than a raw tar | gzip | s3 approach.
  • dryicerx 17y ago
    The --newer switch does the trick
    • whatusername 17y ago
      --newer woks at the file level - I believe that tarsnap works at the block level. So depending on the nature of changes - tarsnap (or anything else block based) would likely still be more efficient.
    • cperciva 17y ago
      Incremental archives via 'tar --newer' are better than just storing complete archives each time, but they fail compared to tarsnap in two important ways: 1. If you modify a file and then perform a backup, you'll store the entire new file rather than only the parts of the file which changed; and 2. If you modify a file and then perform a backup, you now have two copies of that file stored, and you have to continue paying to store both copies even if you only care about the most recent version of the file.
      • dryicerx 17y ago
        Point taken. Of course you are comparing a full personal backup solution meant specifically for this task to a one liner :) But for my purposes as personal data backup, I just wouldn't see a speed or price difference of that type of efficiency.