Does moving files around my computer causes data corruption?

Question

I've got a silly question floating in my head: If I have a file in C:/test1 and move it to C:/test2 then to D:/Other and so on... again and again, may the file become corrupted? What about making copies of the file again and again (copy of copy of copy of copy)?

Answer 1

They ought not to; files are on/off patterns, and all operations such as move and copy are designed to move data whole, without losing any. It's not like the human game of "telephone" where copying the file over and over makes it more mumbled and confused every time until all that's left is nonsense.

But, every time you do something, there's a tiny chance of it going wrong. The more things you do, the more chance you will run into one of them going wrong. One of them, for instance, is a powercut in the middle of a move - what will you end up with? A partial file? No file? Two files?

Nothing is perfect; here, in 2005, a study looks at disk drive error rates, and finds:

We observed 3 loss events while reading 1.4 PB, This is 3 lost files. In the datacenter environment we lost 2 of 32,000 10GB files, In the office setting (System 1) we lost one 10GB file in 35,000 tries, and no 100GB files in 7,560 tries.

That's just disk errors, they also "experienced at least 4 controller firmware or driver errors".

On top of that, there's memory errors - the chance of corruption between loading information into computer memory and then writing it out again. There's an infamous Google study where they track errors in memory accross their thousands of servers, cited at Wikipedia here which states:

"A very large-scale study based on Google's very large number of servers was presented at the SIGMETRICS/Performance’09 conference.[4] The actual error rate [..] 25,000 to 70,000 errors per billion device hours per megabit [..] more than 8% of DIMM memory modules affected by errors per year."

And of course there's a chance of error, a misread, a corruption by power spike or cosmic ray, a design flaw, pushing outside a low tolerance of cheap hardware, in the magnetic media on the disk, in the transfer over cables, in the motherboard, the chipsets, the processor, memory, the firmware or software, and all the way back again.

Incidentally, here is a fantastic article based on the idea that if a computing device corrupts data in memory while looking up an internet address, it might lookup the wrong address. So the author registered some "wrong addresses" (what you'd get if you corrupted a bit of a popular internet site name) to see if any devices connected to them - he had 52,000 requests in almost 6 months.

The study of how to make things /more/ reliable in computing is a busy area. From CRCs and checksums, to RAID with parity, the checksum-tree based filesystem ZFS, to automatic replication, to ECC memory, and so on.

Answer 2

Moving files within the same filesystem shouldn't touch the file's data at all, but only alter the directories containing the file.

Answer 3

Hardware and software can fail at anytime, there are a lot we can do to avoid data corruption, but there is little we can do to ensure there is no data corruption at all. Cosmic ray and even the atmosphere can cause bit rot to the data, how can the majority of users avoid that? Right? Data corruption could also happen during the data transmission, which is the hardest part because 99% of the websites won't provide checksum files for users to test if their downloaded copies are intact.

After all, in my experience, data corruption rarely happens.