Skip to content

Vinicius Grippa

Scott Stroz edited this page Oct 26, 2026 · 2 revisions

← Community Spotlight

1. How did you get started with MySQL, and what was the version you first used?

I started with MySQL when I was a junior DBA at a Brazilian company. That was back in 2014, and the version I worked with was 5.1, while MySQL 5.6 was being released around that time. The good old days when the MyISAM engine was king! :D

2. Who in the MySQL community has influenced you the most, and why?

The first one was Peter Zaitsev, the founder of Percona, and the other was LeFred. They were and still are absolute legends in the community, and they continue to inspire people every single day.

3. What’s one thing the MySQL community has taught you?

First and foremost: be humble. It is insane how many smart people exist in the MySQL community. You will always find someone who has something to teach you if your ears are open to listen!

4. Which MySQL community event or interaction has stayed with you, and why?

FOSDEM for me is the main event of the year. It has a great capacity to bring together members of the community from all over the world. Besides the friendship, it is an event where people share the latest innovations and the most advanced technical sessions in the MySQL ecosystem. It is also a great place to spark ideas for the next cycle of work in MySQL.

5. What’s your favorite MySQL feature that you think more people should know about?

I think MySQL 8 brought many great features, such as dynamic redo log adjustment and the ability to create invisible primary keys. Invisible primary keys allow legacy applications where tables do not have PKs to keep their behavior without needing any code changes, while still improving database performance through primary keys. This is particularly useful for environments where replication is active.

6. If you could add one thing to the MySQL ecosystem tomorrow, what would it be?

Force everyone to use Primary Keys! 😛 Jokes aside, I would love to see MySQL support extensions the way Postgres does, along with better AI support. The good thing is that this is coming fast, and I expect to see both in upcoming releases.

7. What is one thing about MySQL you wish more people understood?

It sounds simple, but it is still commonly misunderstood: how indexes actually work. People assume an index is supposed to fix any performance issue you might have. Because of that, I have seen countless cases where a table had more indexes than columns, even using more disk space for indexes than the data itself. In databases, there is an unwritten rule that every time you improve something, you have to give up on something else. If you create an index, you might gain read performance, but you lose write performance and disk space.

8. What’s something about you that people in the MySQL community might be surprised to learn?

Many people nowadays consider MySQL not to be the right database for modern applications. I think if they saw how active community members are shaping and adding features to MySQL, they would be surprised by how many new things are coming out so fast. There are companies in the ecosystem like Readyset, VillageSQL, and ProxySQL that bring a lot of value to the table. Besides still being the most well known open source database according to DB-Engines, it is definitely a good time to pay attention to the ecosystem.

9. What’s the most interesting MySQL problem you’ve had to solve?

Debugging performance issues is definitely the most interesting problem for me! Collecting perf data, creating flame graphs, and analyzing the source code are really fun to do and can teach you a lot of things. The performance category goes way beyond just creating indexes to fix slow queries!

10. What technology other than MySQL are you excited about right now?

It may sound like following a trend, but definitely AI. It really changed our relationship with technology. The models are becoming smarter and faster, and they truly help people improve productivity and explore new ideas. Plus, this will generate an explosion of data, so I am also excited to see how we will handle all this new data being produced by AI agents.

Clone this wiki locally