Home \ Articles \ Data recovery statistics

Data recovery statistics

ZAR has been discontinued
After about twenty years, I felt ZAR can no longer be updated to match the modern requirements, and I decided to retire it.

ZAR is replaced by Klennet Recovery, my new general-purpose DIY data recovery software.

If you are looking specifically for recovery of image files (like JPEG, CR2, and NEF), take a look at Klennet Carver, a separate video and photo recovery software.
Overview - Data recovery statistics
After analyzing the data kindly reported by ZAR users, we decided that some kind of aggregate data recovery success rate information would be nice to publish. The data also includes some performance measurements (i.e. run speeds). The data is in fact biased due to the limitations of the reporting process. Consider these points The analysis does not include physically damaged devices. One cannot submit a report in case the recovery fails completely (no remains of the filesystem can be found at all). Although there are not many, the totally failed runs are thus not included into the analysis. Although we made a major effort to ensure the validation reflects actual state of the volume, the result is only as good as the validation routines. The report contains data reported by multiple versions of the software (including certain debug versions never released for general use). We actually made some adjustments to the algorithms based on the reported data, and the later versions do thus perform somewhat better. This is not reflected in this report. The analysis does not distinguish between FAT16, FAT32, and the generic FAT filesystem (with tables lost and bitness uncertain). "FAT filesystem" in the following text collectively refers to these three. All the charts are based on about three months worth of raw data.
Filesystems - processing speed

Data recovery speed

This chart illustrates the "simple volume recovery" run speed (gigabytes per minute of the area scanned).

  • The run time is measured from start of the raw data identification to the point when a file and folder tree is displayed. The validation and/or file copy process are not included.

  • The area size to analyze is recorded as provided by user which can be a bit off the actual volume size. This effect is believed not to provide any significant discrepancy.

The slower speeds of the FAT recovery are attributed to the fact that FAT is mostly used on older generations (slower) hard drives and on a slower solid-state devices (e.g. USB flash drives).

Filesystems - data recovery success rate

Data recovery statistics

This is without a doubt the most interesting part of the data. Subject to the limitations described at the beginning of the article (i.e. no physical failures),

  • If a data loss involves a FAT volume, there is an about 50% chance to recover nearly all the data (at least 90% of files). In about 75-80% cases at least 75% of the files can be recovered.

  • If there is an NTFS volume involved, averages are a bit worse: 40% chance for a successful recovery (at least 90% of files) and about 75% chance to recover at least 70% of the files.

Digital image recovery

Digital image recovery processing speed

Because of the design limitations, validation is not used in the digital image recovery mode. Hence we only have the data detailing the run speed. Since the processing overhead for the digital photos is quite low compared to the typical filesystem analysis, this mostly describes the performance of the widespread memory-card-to-PC interfaces. The resulting chart is thus not very exciting, and only provided for the sake of completeness.

Miscellaneous notes
  • This document is based on the dataset as available on 17th August 2006
  • The recovery performed by Zero Assumption Recovery versions 7.8, 7.9 and 8.0.
  • RAID reconstruction stats are excluded from the dataset.
Copyright © 2001 - 2023 Alexey V. Gubin.