Create a large file for testing

create-large-file.sh

# Please note, the commands below will create unreadable files and should be 
# used for testing file size only. If you're looking for something that has
# lines in it, use /dev/urandom instead of /dev/zero. You'll then be able to
# read the number of lines in that file using `wc -l large-file.1mb.txt`

# Create a 1MB file
dd if=/dev/zero of=large-file-1mb.txt count=1024 bs=1024

# Create a 10MB file
dd if=/dev/zero of=large-file-10mb.txt count=1024 bs=10240

# Create a 100MB file
dd if=/dev/zero of=large-file-100mb.txt count=1024 bs=102400

# Create a 1GB file
dd if=/dev/zero of=large-file-1gb.txt count=1024 bs=1048576

# Create a 10GB file
dd if=/dev/zero of=large-file-10gb.txt count=1024 bs=10485760

# Create a 100GB file
dd if=/dev/zero of=large-file-100gb.txt count=1024 bs=104857600

# Create a 1TB file (careful now...)
dd if=/dev/zero of=large-file-1tb.txt count=1024 bs=1073741824

Depending on what your are testing, this may not give you what you want. If the filesystem has inherent and enabled compression (like zfs), these files made with /dev/zero will be compresssed to almost nothing. Use /dev/random in that case.

Nice thanks

Thanks :) As @tandersn mentioned, on a ZFS dataset with compression enabled files written with /dev/zero will compress to almost nothing. But using /dev/random turned out to be really slow. For testing purposes I turned off compression for the ZFS dataset.