I had been doing web dev for most of my coding career. I'd take breaks from it now and then, building games, some small project that looked interesting, or just learning something new. But this time I wanted to go a little deeper.
So I looked around for something interesting and built a simple in-memory key-value store in C. But it felt too simple. It was nothing like the technologies I actually used in web dev, like Redis and databases such as MySQL and Postgres. So I searched around and found the B-tree. I implemented a bit of an in-memory B-tree, then found out a B-tree isn't really what those technologies use either, and found the B+ tree. I finished an in-memory B+ tree. It was messy, but the concept itself was what real databases actually use. So I went looking for how real databases use B+ trees, and that's where I ran into pages, WAL, and so on.
Then I thought about how to add disk saving to it. In the key-value store I had done disk persistence before, just with a CSV file. This time I realized I had to write the disk version from scratch. That wasn't really a problem, I had the general idea of how things worked, I just had to replace the memory pointers with bytes and pages.
