Applies to: Pro, Enterprise, and Audit
Purpose
This assessment combines a complete addressable media read with sequential throughput and latency analysis. It is intended for reuse decisions where read reliability and abnormal performance both matter.
Operation
DriveErase reads the device from beginning to end in 1 MiB blocks and records:
- Unreadable block count;
- Representative bad-block offsets;
- Blocks requiring at least 250 milliseconds to read;
- Representative slow-block offsets;
- Average sequential read throughput;
- Average block-read latency;
- 95th-percentile block-read latency;
- Post-scan SMART information where available;
- Overall health score and grade.
Slow blocks
A slow block is an indicator for review, not conclusive proof of physical damage. Read latency can be affected by:
- Internal error recovery;
- Controller or bridge behaviour;
- Thermal throttling;
- Power-management transitions;
- Background device maintenance;
- Caching;
- HBA or enclosure contention;
- Virtual or shared storage.
Repeatedly slow regions on a directly attached device can support a decision to retire it, especially when SMART or read errors are also present.