Create Random Files For Disk And Internet Testing


If you're trying to sort out a problem with a PC, and especially if it involves trying to diagnose a problem with disk or internet speed, then there's one thing you're going to need pretty quickly: a large file or two, with which to test the system.  You can then try copying that file from one part of the suspect drive to the other, perhaps, or sending it by email to yourself, or whatever other action might help to diagnose the problem.

Being able to create large, random files on demand also has other uses too.  For example, if you want to sell or give away a hard disk, you need to wipe it first.  And the easiest way to wipe it is to overwrite the data on the disk with something that's not so important.  By creating one or more large files of random, harmless data until the disk is pretty much full, you ensure that any important data which was deleted but was still on the disk (such as in the recycle bin) has been overwritten.

I've recently been investigating various programs that claim to be able to create a dummy data file on demand, and the best I found is called Dummy File Creator.  You can download it from and it's free of charge.  It's tiny, at just a 0.02 MB download for the portable version, and it's malware-free according to Virus Total and Web of Trust.

As you can see from the screen shot below, you simply fire up the program, choose the size of the file you want to create (in my tests it created a 20 GB file in less than a minute) and the job is done.  If you're using the dummy files to test disk copying speeds and you want to ensure that there are no easily recognisable patterns in the file that could be cached in order to speed up the results, remember to choose the non-compressible option.  



Please rate this article: 

Your rating: None
Average: 4.2 (10 votes)


A nice utility. However, a file with random content is still compressible. 7-zip compressed a 10MB random content file to 4MB (.7z). The file was uncompressible using the .zip format. Details below.

Generate two 10MB files, "notrandom.txt" and "random.txt." Using 7-zip, compress each file to .zip and .7z formats.

notrandom.txt: 10240KB 13KB
notrandom.7z: 2KB

random.txt: 10240KB 10241KB
random.7z: 4098KB

A good point but the current version reduces the randomness so files can be written more quickly. I would expect some compression utilities to be effective with it as the author points out (see the last sentence that I bolded). It would be interesting if you ran your tests against the more random files from the previous version.

The program only generates a 4MB block of random data which it then rewrites with some bytes randomized again. See the Technical Facts section at the bottom of the web page.

Random Content -
Dummy File Creator 1.2 writes random bytes ranging from 0 to 255. However, unlike previous version which generates true random file content, Dummy File Creator 1.2 uses a different approach in random content generation in order to increase the performance of random content generation. Dummy File Creator now will generate 4MB of random data and reuse the same data by altering only some bytes at random locations for each subsequent write. While the result still defeats all of the compression software we tested (i.e. a larger compressed file than the original size), but it is still possible to compress this pseudo-random content if a specifically designed compression algorithm targeted at Dummy File Creator (very unlikely) is used. To design such algorithm, it must use dictionary words with length ranging from 1 to 4,194,303 bytes. Most people will not notice this change, but for people who are developing compressing algorithms, it is recommended to use the previous version which generates true random contents (but much slower) for testing.

-- There are websites to test internet throughput upload and download speeds like or

-- programs that benchmark disk drive read and write speed & report errors like or

-- Vendor-provided diagnostic tool & low level formatter links

-- generating a checksum pair allows error testing file transfers if your drive has a problem

-- interactively set your DNS (looks up IP)free for best internet performance

Interesting group of links, very useful. The only thing is the third link clearly says 'Vendor-provided diagnostic tools'. There is no way any of those tools can perform a real low-level format, and UserError and Remah have made it very clear in the comments below. Zero-filling, random-filling, and full wipe are not low-level formatting schemes.

I have no idea if the Wikipedia info is correct for all of the possible combinations of the linked drives and software.

But it seems to me that the combination of zero writing to all the drive (less the sectors reserved for manufacturers disk bios etc that used to be on ROM), partitioning and formatting to set disk NTFS / FAT 32 / EXT2/EXT3, JFS, XFS and optimum allocation unit would be the equivalent of a low level format process.

At least that is the entire process set for a low level format that I recall from the days when you had to point to a ROM location to initiate the process. Those steps would seem to check the media, align the tracks, create sector size & count, create the file table that links the sequence AU numbers, etc.


Guess it was a false positive. Puzzling, but it looks safe based on other sources cited.

Hi Rob,

Downloaded the portable (no installer) version of this app.
Unzipped and attempted to execute. Avast blocked execution and gave
me a message that the file was infected with "DS#DynaGen||algo:"
MBAM however gave it clean bill of health.

What do you think?

I guess DS#DynaGen||algo means Datastructures, Dynamically Generated, Algorithm. Which is what the program is all about, dynamically generating random files.

Thanks George.

That sounds rather benign but why would Avast actually block execution of the program and yet give it a clean bill of health per the virus total website? My question is, of course , rhetorical , but it's still a puzzle to me.

that just is one of the many false positives that Avast is so well known for, at least among my critically minded friends.

VirusTotal results doesn't account for on-execution detection of malware and behavioral detection. Certain malware's are caught by the antivirus on execution, after it's extracted and when it reaches the memory. VirusTotal only depends on malware signatures of antivirus.
Both the zip package and .exe scan clean using multiple engines, including that of Avast! MC - Site Manager.
If your purpose is to wipe the harddrive for sale or give away, I would suggest you would be better off using ccleaner with its disk overwrite, say three passes. Mainly because in theory it is possible to get wiped data from what is called "Data remanence" What I used to like about the old eraser programs was they actually wrote "0" on everything, then wrote over the zero "1" and then repeated this cycle several times. It was also a great way of fixing some corrupt floppies or harddrives. The other reason for ccleaner is that parts of the harddrive maybe missed; ccleaner will get it all.

CCleaner is a good option for most people, but I really loved the good old low-level format. Sadly, is almost impossible to do it with modern hard drives, although some free tools claim they can do something very close to the actual thing.

A couple of the HD vendor supplied utilities claim to perform zero-writes on drive - see above

For those who are wondering what UserError is talking about, the Wikipedia article on Disk formatting has a section which explains why nowadays it is usually impossible to do Low-level formatting (LLF) of hard disks. Most of the time when people use the term "low-level formatting" what they are really describing is "reinitializing" which does have the benefit of overwriting all user data.