A blog with no database

This blog has no database. Posts are Markdown files in a folder, the front matter is six lines of key-value pairs, and the whole thing renders through a small PHP front controller. Backing it up means copying a directory.

What a database would have cost#

A CMS brings a schema, migrations, a query layer, a cache, and a security surface that needs patching on someone else's schedule. For a blog that publishes a few times a month, none of that pays rent.

// The entire storage layer, more or less.
foreach (glob(POSTS_PATH . '/*.md') as $file) {
    $posts[] = parse_front_matter(file_get_contents($file));
}

usort($posts, fn ($a, $b) => $b['timestamp'] <=> $a['timestamp']);

Reading forty files per request sounds wasteful until you measure it. The OS page cache holds them all, and the whole listing renders faster than a single unindexed query would.

Where the line is#

Flat files stop being the right answer when you need:

1. Concurrent editors who can conflict
2. Relational queries across content
3. Thousands of items, where scanning a directory turns into real work

None of those are true here, and if one becomes true, Markdown files import into anything.

The parts worth building carefully#

• Front matter parsing — strict, boring, and forgiving of trailing whitespace.
• Slug and filename discipline — the date lives in the filename, the URL does not.
• A real Markdown renderer — including tables, because technical writing needs them.

Everything else is a view.

phparchitecturetooling