This month in MariaDB Foundation: Feb 2025

Adoption Index 1

The act of measuring disturbs the system and changes its state.” is a quote attributed to Werner Heisenberg. While I am uncertain about whether Heisenberg ever uttered those exact words, I am certain that the quote “What doesn’t get measured, doesn’t get managed” is a management adage that gets a lot of negative publicity, deservedly so.

Metrics Not Considered Harmful

Yet, metrics can be helpful. MariaDB Foundation is far from overmeasuring pointy-haired-boss numbers, and in February, we made an attempt at systematically creating a meaningful metric that is easy to calculate.

One of the drawbacks of Open Source is that we have no self-evident metric of our community size. How many users does MariaDB Server have? That’s an easy question, and it’s hard to answer. What exactly is a user? We don’t require registration, and we always want to give priority to growing the ecosystem over getting more exact measurements of its size.

Enter MariaDB Adoption Index (Beta)

Uncertainties are challenges, not reasons to give up.

So now, we are for the first time sharing a monthly public MariaDB adoption index. Why? We want to establish easily accessible numbers on how MariaDB [Foundation] is succeeding. 

As this is a new concept, you will have to forgive me for being a bit wordy when introducing the concept. For any TL;DR fans amongst the readers: Look at the graph, glance at the table, and move to the section “Back to our Six Goals for 2025”.

For this initial month, we have put together an index that summarizes 21 KPIs spanning from actual downloads on mariadb.org to social media followers.  We will strive for consistency and flexibility: KPIs and weights may be tweaked in the following months based on good feedback and reasoning, but the main idea will stay the same.

That idea is similar to share indexes. We create a number, define it to be 100 at the beginning of January 2024, and then let the KPIs – much like individual stocks – influence the index over time. New “shares” (KPIs) can be included, but the retroactive values should stay the same.

Adoption Index 2
MariaDB Adoption Index – beta version Feb 2025

In the above picture, the index has grown from 100 in 2024-01 (not shown, as we show the rolling 12 months) to 145 in 2025-02.

However, we still label this a “beta version” of the index. The computation of the index isn’t written in stone, as the 21 KPIs we chose are not the only possible ones, nor is their internal weighting (“the Market Cap of the individual KPIs”) obvious.

Hence, consider the current index a “RFC”, a request for comments.

How we computed the MariaDB Adoption Index

Here are our KPIs in some detail:

Adoption Index 3
How we computed the MariaDB Adoption Index

The weighting of the individual KPIs is shown in the pie chart on top. Here are a few thoughts on how we chose them:

  • The adoption category is 40% in total: The most genuine indicator would be the amount of MariaDB databases running. As such a number is not possible, the next best thing we have identified are downloads from mariadb.org (20%), installations through Debian (10%) and the amount of docker official image pulls (10%). Unfortunately Docker image pull statistics have in general gone through a major abnormality from the end of 2023 to September 2024, so we decided to include Docker pulls from January 2025 forward with a neutral impact on the previous months. 
  • Contributions have in total a 19% weight. The amount of contributions is an important indicator of an open source project’s health – so a major indicator in the index is the amount PRs per month created by external contributors (5%) and the amount of unique contributors (5%). In addition we are following general popularity with GitHub stars (3%), mention of MariaDB in other GitHub projects’ READMEs (3%), as well as new users on MariaDB Zulip chat (3%). 
  • The external category accounts for 20% and how well MariaDB is doing on Google Trends (5%), in Wikipedia article views on all languages (5%)  as well as on the DB Engines ranking score (10%) that encompasses several metrics in itself. 
  • The developer network at 9% represents popular forums for developers: Reddit, Hackernews and Stackexchange where discussions about MariaDB indicate developer activity with MariaDB. 
  • Social Media has in total a 12% weight and gives an overview of how well posts about MariaDB are reaching developers, decision makers, users and other interested stakeholders. We are following followers/subscribers on LinkedIn, Youtube, Fosstodon, Instagram and X. 

Challenges when defining the MariaDB Adoption Index

We have strived to use only numbers that are publicly accessible. For example, YouTube content views are only accessible by the account administrator whereas YouTube channel followers is a public number.  

There are a lot of seemingly trivial choices to be made with KPIs. Do we follow deltas per month or cumulative numbers? The amount of followers/subscribers instead of actual impressions or views? We have tried to make sensible choices. 

There are challenges with some seemingly basic numbers, like the amount of Docker official image pulls stats that we realized have major abnormalities for MariaDB (and MySQL and PostgreSQL). During the time period from October 2023 to October 2024, the amount of pulls per month show x4 or even x5 amounts due to technical changes at Docker that triggered automatic repulls of images. After trying to normalise the numbers, we decided on starting Docker pulls in January 2025 in the index. 

So, how well have we done? What should be tweaked or added? Let us know!

Future plans: External reproducibility

We have been tipped about the Scarf solution that helps gather sales and marketing intelligence for open source. We’re looking into how to apply that on MariaDB.

And while we are still iterating our definitions, most of the work has been manual and/or based on Google Spreadsheets. Over time, we expect to automate the data collection and make it more public. Think a GitHub repository, curl, json objects, and a few GitHub Actions. Stay tuned!

Back to our Six Goals for 2025

You might remember the Six Goals for 2025, that I introduced in our January update.

  1. Community Analysis: Know where we stand!
  2. Contributions: Make MariaDB vibrant as a developer community!
  3. Outreach: Evangelise the current and potential users to drive adoption!
  4. Vectors & AI: Bring in new users to MariaDB Server!
  5. Enable MySQL users to use MariaDB even more easily!
  6. MariaDB for Universities Program: Support database tuition with MariaDB!

While we were busy working on all of them in February, I want to highlight on the most visible outcome – our work in Outreach.

MariaDB Day during FOSDEM in Brussels

MariaDB Day during FOSDEM in Brussels was a success, as described in the blog entry MariaDB Day Brussels Aftertaste 2025-02. Some details:

  1. Nikita Malyavin, MariaDB / Sergei Golubchik, MariaDB, When a client and a server are just a PARSEC apart
  2. Vicențiu Ciorbaru, MariaDB Foundation, Launch Your Open Source Career: First Steps in Contributing to MariaDB
  3. Sergei Golubchik, MariaDB, Meet the newest MariaDB releases
  4. Otto Kekalainen, Server-initiated instant failover in MariaDB
  5. Peter Zaitsev, Percona, Splendidly wonderful features in MariaDB 11
  6. Kristian Nielsen, MariaDB Foundation, MariaDB replication: Ease-to-use or easy-to-abuse?
  7. Roman Nozdrin, MariaDB, Ascend to New Heights: Discover What ColumnStore Can Really Do for You
  8. Sergei Golubchik, MariaDB, Get to know MariaDB Rocket-Fast Native Vector Search
  9. Jags Ramnarayan, SkySQL, Building GenAI-Powered Apps with MariaDB’s VectorDB and SkySQL’s AI Agents
  10. Diego Dupin, MariaDB, Firing Up the Boosters: concrete cases of using an LLM with MariaDB

Several of these were related to other goals – to contributions, to our Vector flagship feature of MariaDB 11.8, and to enabling MySQL users to use MariaDB even more easily.

After FOSDEM, I held a talk on open data at State of Open Con25 in London: Saving the world by feeding the Al Dragon with unbiased open data. The main MariaDB angle here is MariaDB Vector as the tool for feeding the overall AI dragon. While the header is a tad dramatic, so is the state of the world – if you have a quarter of an hour to spare, you can listen to a recording of the presentation, pasted together with the slide deck into a YouTube video linked from the blog entry above.

New releases – and an LTS blog

February is a release month. Ian Gilfillan wrote two release announcements

On a related note, we shared the news that MariaDB 11.8 is LTS – becoming GA in May.

External observations

To conclude, here are a few noteworthy external observations on MariaDB, from Reddit, Medium and elsewhere:

What do you think?

What do you think? Do share your feedback with us! Over comments, Zulip, or email.