Benchmark databases in Docker: MySQL, PostgreSQL, SQL Server

Plus MariaDB, Percona and TimescaleDB

Why should you do database benchmark?

I was a long-term SQL Server user. But it was not my decision to use SQL Server. I was young and had to listen older colleagues. If they said, SQL Server is the best, then it was.

Then I was a long-term MySQL user. But it was not my decision to use MySQL. I was older, but my other colleagues were much more experienced. If they said, MySQL is the best, then it was.

I never searched for any other opinions, because at that time, I was learning. How can I argue with my much more experienced colleagues?

Then I found MariaDB. Then I found Percona. And I asked my much more experienced colleagues something like: “Hey, what are they? Why are they?”. The answer was almost always something like: “MySQL is the best, I know MySQL. Do you suppose to use something else? Do you want to break things?”

And of course, I didn’t want to break things.

But curiosity is my second name, so I decided to find out, which database should suits the best for our application. And I hit the perfect time, because Docker was arriving to the scene, so it looked like, it would be easy to start my work.

So, to answer the question “Why should you do database benchmark?” is to say: “Because it turns YOU into that much more experienced colleague”.

How I did my database benchmark?

To benchmark a database can be a tricky thing. Just look around the internet. There are different benchmarks, measuring different things, sometimes parameters, that do not tell you anything meaningful.

From my perspective, I see database as something, that exactly matches its name. A base for a data. Nothing more. So no application logic in database. Database should hold the data and make two main operations as quick as possible: read and write.

I see read as something not changing the database and write as something changing the database. For me, delete and update are both subset of write.