Monty Taylor d496974e5e Add a spec for creating an Apache Iceberg storage engine
Rather than making our own "cold storage in object storage" engine
like Maria did, we should leverage Iceberg, since it's already used
in a ton of places. Would be a really cool interop story.

Change-Id: I3e94256f7d9c63d9a8ab7fe8566f61cdc26ba9a5
2026-07-13 16:32:37 +00:00
2026-06-07 00:20:17 +00:00
2026-06-18 00:26:43 +00:00
2026-06-20 23:52:47 +00:00
2026-07-01 16:06:37 +00:00
2026-06-18 00:26:43 +00:00
2013-04-22 13:29:49 -07:00
2026-06-19 00:17:49 +00:00
2010-09-28 00:45:44 -07:00
2026-05-18 17:34:22 -07:00
2026-06-20 19:56:10 +00:00
2008-07-19 22:41:52 -07:00
2026-06-30 22:28:26 +00:00

Drizzle

A Lightweight SQL Database for Cloud and Web

What, this again?

Drizzle is a community-driven open source project that is forked from the popular MySQL database.

The Drizzle team has removed non-essential code, re-factored the remaining code and modernized the code base moving to C++.

Charter

  • A database optimized for Cloud infrastructure and Web applications
  • Design for massive concurrency on modern multi-cpu architecture
  • Optimize memory for increased performance and parallelism
  • Open source, open community, open design
  • Container-first approach

Scope

  • Re-designed modular architecture providing plugins with defined APIs
  • Simple design for ease of use and administration
  • Reliable, ACID transactional

Agents

Read https://sfconservancy.org/llm-gen-ai/llm-backed-generative-ai-recommendations.html

Drizzle has been GPLv2 for a VERY LONG TIME. Therefore, as per points 9 and 10 of the above, regardless of the specific copyrightability of the output of an LLM, the safest view is that everything here is a derived work and thus is GPL. Long live the four opens.

  • Drizzle welcomes Agentic development
  • Drizzle is using Agentic code review
  • Drizzle is aggressively gating with Zuul to protect itself from its new love of agents
  • Agents can produce huge volumes of code. Humans cannot read huge volumes of code.
  • Drizzle rules apply whether it's an agent or human:
    • Small reviewable patches
    • More than one day of human effort means the patch is too big. Start over. Break it up.
    • Assume it's being reviewed by someone with ADHD
  • OpenStack rules apply whether it's an agent or human:
    • If it's not tested, it's broken

Compiling from source

Just build the container:

podman build .

Yes, docker works too, don't be silly. But I love rootless podman myself. All of Drizzle's gate jobs use rootless podman.

Dependencies

The dependencies are listed in bindep.txt to be used with either bindep or bindep-rs. The Containerfile uses bindep-rs to avoid polluting the dependency list.

Compiling

autoreconf -i
./configure && make && make test

But seriously, just build the container. The chances this builds TODAY on your laptop are pretty much none, as we're still forward-porting from Ubuntu Precise.

Running Drizzle

Just run the container!

Fun story - every config value in Drizzle is settable via env vars which is how all the fancy container people like it. Been that way since 2008. What's old is new again.

Cheers!

  • The Drizzle team
S
Description
Drizzle Database Service
Readme 108 MiB
Languages
C++ 85.9%
C 8.1%
M4 2.7%
Shell 0.7%
Roff 0.7%
Other 1.6%