dd can create random files of any size

Creating a 100MB file full of random binary data

dd if=/dev/urandom \
    of=a_file \
    bs=1048576 \
    count=100

dd

A command-line utility for copying and converting data

if

the input file

/dev/urandom

takes advantage of UNIX's philosophy of treating all resources as files. This is a standard UNIX utility (it generates random bytes), but to dd it works as if it was a file

of

the output file

a_file

in this case we generate an actual file

bs

block size; the amount of data to read/write at once. It should be a power of 2

1048576

That is the closest power of 2 to 1 MB

count

How many blocks to write

100

We are creating a 100MB file full of gibberish

Creating a 1Kb file full of random ASCII data

The same command can be adapted to create ASCII data, by piping another command that translates the output to ASCII

dd if=/dev/urandom \
    bs=1 \
    count=1024 \
    2>/dev/null \
    | base64 \
    | head -c 1024 \
    > a_file.txt

dd if=/dev/urandom

The same command as above (except that there is no output file this time)

bs=1 count=1024

this time just 1 byte of data at the time, for a total of 1024 bytes

2>/dev/null

if there are any errors or debugging info, just send them to the "black hole" (/dev/null) so we don't see them

| base64

send the output to the util `base64`, which converts binary data into text. It is commonly used to encode, say, images in email attachments

| head

send the output to the util `head`, which takes only the initial part of any ouput and discards the rest. This is needed, because base64 makes the output larger by about 33% (there is a reason why we use binary and not text to encode data)

-c 1024

in this case only take the first kilobyte (note how it matches the initial count)