mysqldumpsplit

Split a mysqldump into separate files for each table.

Usage of mysqldumpsplit:
  -combine
      Combine all tables into a single file, deletes individual table files
  -combineFile string
      The path to output a single SQL file
  Only used if combine flag is set (default "dumpfile.sql")
  -i string
      The file to read from, can be a gzip file
  -o string
      The output path  (default "output")
  -skipData value
      Comma separated list of tables you want to skip outputting the data for.
  Use '*' to skip all.
  -skipTable value
      Comma separated list of tables to skip.
  Names can contain '*' for wildcard values
  -version
      Display the version and exit

Install

If you are on a Mac you can install with Homebrew.

brew install afrase/tap/mysqldumpsplit

You could also grab the binary from the releases page.

Performance

Using a sort of pipelined approach, mysqldumpsplit is very memory efficient. The amount of memory used will mostly be based on how large each line in the file is.

I was able to split a 130 table 16GB sql file in less than 30 seconds and use less than 20MB of RAM.

Usage

Split a SQL dump into individual tables.

./mysqldumpsplit -i ~/Downloads/dump.sql -o ~/Downloads/output

Skip the data for specific tables.

./mysqldumpsplit -i ~/Downloads/dump.sql -o ~/Downloads/output -skipData table1,table2,table3

Combine all tables back into a single file. Deletes the output directory once finished unless the combined file is in the same directory.

./mysqldumpsplit -i ~/Downloads/dump.sql.gz -o ~/Downloads/output -skipData table1,table2,table3 -combine -combineFile ~/Downloads/dumpfile.sql

Skip data for all tables.

./mysqldumpsplit -i ~/Downloads/dump.sql.gz -o ~/Downloads/output -skipData *

Know issues

• If everything in the file is on a single line this will not work.
• Even if skipping the data for a table, it still must be read by the program.