Soft corruption in data storage implies that the data retrieved from a storage device differs from the data written to that device. The corrupted data can exist indefinitely on storage devices. You might not discover this corrupted data until you retrieve and attempt to use this data.

Depending on the configuration type, a Redundant Array of Independent Disks (RAID) logical volume (LV) prevents data loss when a device fails. If a device consisting of a RAID array fails, the data can be recovered from other devices in that RAID LV. However, a RAID configuration does not ensure the integrity of the data itself. Soft corruption, silent corruption, soft errors, and silent errors describe data that has become corrupted. This occurs even if the system design and software continues to function as expected.

When creating a RAID LV with Device Mapper (DM) integrity or adding it to an existing RAID LV, consider the following points:

• The integrity metadata requires additional storage space. For each RAID image, every 500MB data requires 4MB of additional storage space because of the checksums added to the data.
• While some RAID configurations are impacted more than others, adding DM integrity impacts performance due to latency when accessing the data. A RAID1 configuration typically offers better performance than RAID5 or its variants.
• The RAID integrity block size also impacts performance. Configuring a larger RAID integrity block size offers better performance. However, a smaller RAID integrity block size offers greater backward compatibility.
• There are two integrity modes available: bitmap or journal. The bitmap integrity mode typically offers better performance than journal mode.