Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

When you have corrupted data: through packets, or hard drives, what kind of errors do you usually find?

Is it a bit tranposition (switched bits)?

Is it random noise? (flipped bits)?

And when you do get an error, how much corruption do you usually see? (just 1 or 2 bits or a whole block).

Or do you see any other patterns.

share|improve this question

4 Answers 4

up vote 2 down vote accepted

I've seen all your examples, sometimes several at once. But I wouldn't use the term 'patterns', corruption really can occur anywhere in a file and look like anything you can imagine.

share|improve this answer
    
+1: That's exactly what I want to know. If it more often random, or more often an observable pattern. –  Unknown Jun 9 '09 at 7:25
    
the time to worry is when its undetectable. This could happen in a few contrived circumstances - I've seen data from a seismic survey that was corrupt but not obviously so, it was only discovered when the field crew brought the tapes in and we tried to process the data. Luckily we traced the fault and figured out what it was doing and (with an idea of what the corruption looked like and what the data "should" look like) managed to work around it... would have been massively expensive to re-shoot the data in the field. –  RobM Jun 9 '09 at 10:45

I don't think this question has a real answer, unless someone's done statistical analysis on which type of error is likely to occur. It could be anything.

share|improve this answer
    
Well, I couldn't find any papers on it. So I am asking you all who have very likely dealt with corruption on the field. –  Unknown Jun 9 '09 at 7:00

On one hardware project I worked on we were using a milled PCB instead of a properly grounded one. When we started to see corruption at higher signal rates caused by cross-talk on the data lines. It typically showed up as one or more bits set high next to other legitimately set bits.

I would also guess that other systems (e.g. hard drives) have thier own properties that effect how data my be corrupted.

share|improve this answer

Corruption implies random damage to your data. Corruption doesn't usually look like anything -- not in the way that you're implying by your question.

As Artem has already answered, there really isn't a definitive answer. It all depends on what your system is doing and what it's meant to do.

Basically, you need to learn what is normal for your system. When your system starts exhibiting behaviour outside of normal, you need to look for the root cause of the problem. One of the causes you should consider is data corruption.

I've seen data corruption manifest in a number of ways, including:

  • Significant increase in input validation errors for batch processes.

  • Reports with unexpected results. ie the values are out of the expected range.

  • Dial-up sessions consuming gigabytes of data in a matter of minutes. ie the data throughput was physically impossible.

  • Customers being billed stupidly large amounts of money. The reverse is true too -- stupidly large credits!

  • Programme crashes because the number does not fit into a two byte integer.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.