vision dead 2008

MySQL 6.0 Drizzle

'Smaller, slimmer, faster MySQL.' Backed by Google, Sun, Rackspace, Intel, HP. Discontinued anyway. Patron list ≠ contributors.

What it is

Drizzle was a relational database management system forked from MySQL 6.0's development branch. It used a microkernel architecture written in C++ where most functionality was provided through plugins. It deliberately removed MySQL features deemed unnecessary for cloud/web workloads: stored procedures, views, triggers, the query cache, and the complex ACL system. It targeted multi-core architectures and web-scale applications.

The story

Drizzle was the cool kid's MySQL — a radical reimagining of what a relational database should be in the cloud era. Started in mid-2008 by Brian Aker, a senior MySQL architect at Sun Microsystems, Drizzle's thesis was that MySQL had become bloated and enterprise-focused when the future was lightweight, cloud-native microservices. Strip out the stored procedures, the views, the triggers, the ACL system, and half the other enterprise features, and what you'd have left would be a lean, mean, web-scale machine.

The timing seemed perfect. Cloud computing was exploding, and the idea of a 'micro-kernel database' with a plugin architecture resonated with developers building the next generation of web applications. The project attracted an absurdly impressive roster of corporate sponsors: Google, Sun/Oracle, Rackspace, Intel, Hewlett-Packard, Red Hat, Canonical, Percona, and others. If blue-chip sponsor lists were a currency, Drizzle would have been a unicorn.

But sponsor logos on a website don't write code. Despite the corporate backing, Drizzle struggled with consistent contributor commitment. Many sponsors contributed developers part-time or for specific features, but the sustained, dedicated development effort a database engine requires never fully materialized. The first GA release didn't arrive until March 2011 — nearly three years after the project started — and by then the landscape had shifted.

NoSQL databases like MongoDB and Redis had captured the 'cloud-native' mindset that Drizzle was targeting. MariaDB had absorbed the 'better MySQL' narrative. And MySQL itself, under Oracle's stewardship, had actually improved significantly with MySQL 5.6 and 5.7. Drizzle was squeezed from every direction.

Development wound down around 2013-2014, and in July 2016, the maintainers officially pulled the plug, stating simply that 'none of us have any time to dedicate to Drizzle anymore.' It was a quiet death for a project that had once promised to revolutionize databases. The patron list had been impressive; the contributor list, less so.

Timeline

Brian Aker begins Drizzle as a fork of MySQL 6.0 at Sun Microsystems

Drizzle publicly announced; attracts major corporate sponsors

Development continues with contributions from Google, Rackspace, and others

Drizzle 7.0 (first GA release) finally ships after 3 years of development

Development activity begins declining as contributors shift focus

Active development largely ceases

Maintainers officially end the project, acknowledging no one has time to continue

Key people

Brian Aker
Drizzle creator, former MySQL Director of Architecture at Sun
Monty Taylor
Core Drizzle developer, infrastructure architect
Stewart Smith
Canonical developer, major Drizzle contributor
Andrew Hutchings
Core developer, SkySQL/MariaDB engineer

Impact

Drizzle's most lasting impact was as a cautionary tale. It proved that having a brilliant vision and an all-star sponsor list means nothing without sustained contributor commitment. The 'strip MySQL down to its essentials' philosophy was ahead of its time in some ways — the microservices database concept would later be partially realized by projects like CockroachDB, TiDB, and Vitess — but Drizzle itself never reached the critical mass needed to sustain an independent database engine.

Several ideas from Drizzle — particularly its plugin architecture and its emphasis on removing legacy cruft — influenced thinking in the broader MySQL ecosystem. Some Drizzle contributors went on to do significant work on MariaDB and Percona Server. The project's real legacy is intellectual rather than code-based.

Lesson: A long list of corporate sponsors does not equal a long list of committed contributors — logos don't write code.

Related forks