Data corruption

Content sourced from Wikipedia, licensed under CC BY-SA 3.0.

Data corruption is when computer data is accidentally changed while it is written, stored, read, or sent. It can cause small errors or make a file unreadable, and in worst cases, crash the system.

Where corruption happens
- It can occur anywhere: in memory, on disks, during transmission, or while processing.
- Some malware can intentionally corrupt files; a computer crash can result if critical files are affected.

Types of corruption
- Silent (undetected) corruption: there are no obvious signs, so the wrong data can go unnoticed.
- Detected corruption: errors are noticed, but they may or may not be fixed automatically.

Causes of data corruption
- Hardware failures (bad disks, memory issues, power problems, aging devices).
- Software bugs and crashes.
- Transmission problems (network or wireless interference, interruptions).
- Environmental factors (cosmic rays, vibration, loose cables, interference from other devices).
- In some cases, corruption is discovered only after the fact, during data checks or scrubbing.

Why it matters
- A corrupted file can be partly damaged, completely unusable, or cause other parts of the system to fail.
- Silent corruption is especially risky because it can spread damage without warning.

How we detect and fix it
- Checksums and error-correcting codes (ECC) help find and fix many errors.
- RAID systems with parity can reconstruct lost data after a disk failure.
- Some file systems (like ZFS, Btrfs, ReFS, HAMMER) keep internal checks to catch corruption.
- Data scrubbing regularly checks and fixes data before errors pile up.
- Disk drives often remap bad sectors automatically; SMART tools monitor drive health.

What you can do to reduce risk
- Use redundancy and backups so you can restore clean data if corruption is found.
- Employ end-to-end data protection that checks data across different system layers.
- Enable and monitor ECC memory, checksums, and data scrubbing on storage.
- Keep firmware up to date and watch disk health with SMART or similar tools.

If corruption is detected
- Restore from a clean backup, retry transmissions, or re-read the data.
- Use redundancy to reconstruct damaged data and prevent recurrence.