Date: Tue, 30 Sep 2008 21:02:35 +1300

From: pgut001@cs.auckland.ac.nz (Peter Gutmann)

Subject: Risks of all-encompassing backups

With users squirreling their data away in ever more obscure locations (this "disk drive" is an iPod, that "disk drive" is a cellphone, the other "disk drive" is an SD card, ...) it's necessary for backup software to be very methodical in what it backs up or face the risk of losing user data.  So what happens when your software to uses a comprehensive backup policy?

Here's one example, with identifying marks deleted:

  This programme, always running in the background, monitors files on your

  computer and notices when they have been modified.  It then copies the

  files, compresses and encrypts them, and sends them through the net to a

  backup computer.  This system reads and preserves ALL FILES on each

  computer.  Users are not allowed to restrict files from being read and

  backed up.

  If you have a laptop, you may have noticed that this programme uses huge

  amounts of bandwidth initially, because it starts out by dumping all the

  files on your disk.  I discovered this when most of my ISP monthly

  allocation was used up over one weekend, largely by this backup.  I

  quickly learned to put the application on "pause" whenever it was plugged

  in at home.  I don't always remember to "unpause" it when I am at work,

  and I'm having second thoughts about whether I even want to.

  After some consultation, I was assured that the bandwidth for uploading

  files would decline rapidly once all the files had been transferred, but

  the high upload rate continued for over a month.  I was mystified why it

  should be taking so long to finish this initial task for an 80GB drive

  until I discovered that the programme is not simply monitoring the

  internal hard disk, but all memory devices accessible to the computer.  So

  when I took it home, it was, among other things, backing up the 300MB

  drive I use for family and personal matters, and another 500MB drive that

  I used as a "hot backup".  In fact, apparently, every time you drop a CD

  or DVD into a drive, or connect a memory stick, it also grabs those files

  and uploads them.  Even connecting a camera, apparently will result in

  your pictures being uploaded and saved.  I haven't yet been able to

  determine whether it is also accessing remote disks that are available to

  my computer at home through my network behind a firewall, where sharing is

  wide open, and other members of my family have information they definitely

  do not want uploaded.

It seems the vendors are stuck between a rock and a hard place.  If they miss some obscure storage location, then customers get upset.  But if they do scour every piece of storage media, then other customers get upset.  You can't even exclude "obvious" media like CDs/DVDs because with packet-writing software you don't know whether what's in there isn't being used as general R/W data storage and therefore in need of backup.