Let's face it — we have all had that dreaded moment when Windows screams almost full disk occupancy. You try Explorer, Storage Sense, and other existing tools, but nothing gives you a clear answer. Sometimes you get a cluttered visual experience, though. So, how do you get an easy answer to what's hogging your disk space?
WizTree is one of the well-known disk scanning solutions for Windows. Its ingenuity is in reading the Master File Table (MFT) directly from the disk, resulting in fast scan times. This, however, comes at the cost of having elevated (administrator) privileges to run it. On the other hand, if you like Open-Source, WinDirStat is a popular alternative that allows you to scan the disk either in admin or user mode.
Now, the question that you should ask yourself is: is disk scanning speed the primary success metric for you? Or is it a secondary aspect to the key utility of having meaningful disk scan results and actionable insights? Moreover, do you want your disk scanning process to occupy a bulk of your system memory, rendering your perhaps already slow system even slower?
Personally, I have often been frustrated by disk scanning software. And why do not we have the simplicity of ncdu? My own frustrations — and a desire to learn Rust and Tauri — led me to build Drive Sonar — a blazing-fast, unbelievably light, user-mode disk usage analyzer that is also cross-platform.
So, what can it do? Drive Sonar:
- scans disk occupancy and usage at a given location
- does a directory listing, which can be navigated
- shows the usage as a treemap — a proportionate visual representation
- plots a histogram of file types and sizes
- draws a size vs. age scatter plot to identify potentially older and large files
- allows printing the scan report as a text file
- opens file explorer in-app and allows deleting files post confirmation
Although Drive Sonar is optimized for Windows, it is cross-platform and runs on Linux and macOS as well. But don't trust my words about its performance; trust the numbers instead. Experimental results, summarized in the table below, indicate that Drive Sonar consumes about 1/6th of the memory when running a disk scan as compared to one of its alternatives. The cold scan time is about 61% of a contemporary solution. On the other hand, the warm scan time is just about half of that software!
| Metric | Drive Sonar | WinDirStat | WizTree |
|---|---|---|---|
| Version tested | 0.6.0 | 2.8.0 | 4.3.1 |
| License | Apache 2.0 | GPL v2 | Proprietary |
| Runs in user mode (no admin required) | ✅ | ✅ | ❌[1] |
| Installed size [MB] | 5.7 | 7.1 | 20.0 |
| Peak memory usage (median)[2] [MB] | 108.7 | 239.6 | 635.2 |
| Warm scan time (median)[3] [s] | 3.3 | 9.1 | 6.5 |
| Cold scan time (single run)[4] [s] | 10.6 | 22.9 | 17.2 |
| Runs on [5] | Windows, Linux, macOS | Windows only | Windows only |
Of course, disk scanning times can often be skewed. WizTree's MFT-read trick can be really fast in some cases, particularly when a cold scan is considered. On the other hand, WinDirStat is a long-trusted default in the Open-Source domain.
Drive Sonar is a new entrant to this field, aiming to provide fast, lightweight scanning, meaningful insights, and a clean user experience. If you don't mind running disk scans with admin privileges and want to make use of fast MFT reads, try WizTree. If you like Open-Source and want to have the flexibility of running disk scans in either user or admin mode, try WinDirStat. However, if you prefer clean, meaningful insights and don't mind trying new software, Drive Sonar is here to help.
Drive Sonar is an Open-Source (Apache-2.0 license) solution. Cross-platform binaries (unsigned) are available for different architectures. Drive Sonar will soon also be available as a Snap package for Linux.
What do you think about Drive Sonar? How was your disk usage scanning experience? Did you print a scan report? Let me know in the comments!
Top comments (2)
Appreciate the “is speed the only metric?” framing, and the user-mode vs admin/MFT tradeoff is the part most writeups skip.
I’ve been working on Oxide on the MFT + treemap side for Windows, with a hard read-only rule (no in-app delete) and more attention to settled memory after the scan than peak during it. Different methodology than your table, so I’m not dropping numbers here and muddying yours.
Curious how you think about peak vs settled RAM, and whether delete-in-app belongs in an analyzer or should stay out. oxidedisk.com if you ever want another data point for a follow-up.
Thanks for the insightful comments. Just checked out Oxide -- the website looks great.
I haven't really considered the settle-time memory usage of Drive Sonar or the others yet. I think this could be a useful metric. However, I have noticed that the peak memory usage stays roughly the same for 2-3 seconds after the scan is over. So, perhaps need to define what "settle time" is before measuring memory usage.
Regarding delete-in-app, I wouldn't say this is a heavily used feature. Although, at times when I use
ncdu, I wish I could delete a file or two from there. That is why Drive Sonar allows file deletion, but only after confirmation. I think I'd like this confirmation to be even stricter in the future.