InnoDB is the most popular transactional storage engine for MySQL. Unlike MyISAM, InnoDB is not only ACID compliant, but it also supports row-level locking and referential integrity. Likewise, instead of employing locking to manage concurrency, InnoDB employs multi-version concurrency control; thereby eliminating the issues commonly caused by lock-based database architectures in large-scale transactional environments.

History

InnoDB was created by Heikki Tuuri on January 20th, 1994. It's original design goal was to be the fastest disk-based relational database management system. As such, many of InnoDB's architectural decisions and algorithms are based on the well-known research of Jim Gray and Andreas Reuter as found in their book, Transaction Processing: Concepts and Techniques. In regard to InnoDB's MVCC design, it similar to the Oracle Database multi-version read consistency model in that it rebuilds a row-version based on UNDO log data.

InnoDB was originally distributed with MySQL 3.23.34a on March 12th, 2001.

Features

InnoDB supports the following features:

• ACID-Compliance
• Storage Management
• Referential Integrity
• Row-level Locking
• Hot Backup
• Multi-version Concurrency Control
• Compressed Tables
• Two-phase Commit
• Asynchronous I/O

Major Users

InnoDB has successfully been deployed and heavily used at flikr, Google, Wikipedia, Paycom, and others. It has been deployed on systems storing over a terabyte of data.

License

InnoDB is licensed under the GNU General Public License (GPL).

Trivia

While InnoDB is a storage engine for MySQL, it is actually an almost complete database in its own right. Though several features are unused by MySQL, InnoDB contains its own parser, buffer manager, and primivitve client/server communication system. However, rather than compete against MySQL, InnoDB was able to fill-in the transactional, ACID-compliant gap MySQL had.

Links

• InnoDB Website