For any application that works with a lot of data, the obvious approach is to store it in some kind of database. Depending on your ambitions, that can be something as modest as SQLite, or as upscale as PostgreSQL.
What often complicates things is not which database to use, but how to use it. Enterprise-scale databases like PostgreSQL give you great power but impose an infrastructure tax: Setting up and configuring a database server is nobody’s idea of fun. Docker containers are an option, but if you’re not already invested in…
For any application that works with a lot of data, the obvious approach is to store it in some kind of database. Depending on your ambitions, that can be something as modest as SQLite, or as upscale as PostgreSQL.
What often complicates things is not which database to use, but how to use it. Enterprise-scale databases like PostgreSQL give you great power but impose an infrastructure tax: Setting up and configuring a database server is nobody’s idea of fun. Docker containers are an option, but if you’re not already invested in Docker as a solution, moving everything to containers is an outsized project.
One reaction to all this has been the rise of self-contained database solutions. Everything is packed into a single executable or living in a single directory, with no external dependencies and no formal installation. SQLite is the obvious example of this style, but it’s not always the most robust solution. There are more powerful databases that do the same thing—including some familiar A-list options. Even better, some come as drop-in application libraries for popular languages.
MariaDB
If you want a fully self-contained instance of MariaDB, you have three choices: Create one yourself with a little work; use a third-party pre-packaged version; or use a portable full-stack solution that includes a web server and other tools along with the database.
Nathaniel Sabanski has outlined the steps required to manually set up MariaDB in a standalone edition. Download and unpack the binaries, create a basic configuration file, then start the server and any management tools you’d need along with it. Any upgrades or maintenance are entirely on you, but the gist describes how to get it done without too much sweat. Note that if you’re using Microsoft Windows, get the .ZIP file to make it easier to extract, and use the mysql_install_db.exe application to set things up (the defaults should be fine).
For even more convenient downloading and unpacking, Andy Targino provides self-contained versions of the MariaDB binaries.
If you need to stand up a web server along with the database, and maybe a few other components, too, look to the XAMPP stack. This all-in-one solution contains MariaDB plus the Apache web server, the PHP runtime, the Mercury SMTP mail server, web-based controls for all the components, and a service manager for the desktop. It even includes OpenSSL for proper https support.
PostgreSQL
Various repackagings of PostgreSQL as a standalone application have come and gone over the years (see this project, for instance), but it takes relatively little work to set up your own standalone PostgreSQL application. Obtain the binaries minus the setup tools, unpack them into a directory, and run initdb to configure the basic setup. You can then use pg_ctl to start and stop the database as needed.
Python developers have a truly slick option for adding a self-contained PostgreSQL instance to an application: pgserver, a pip install-able library that contains a fully standalone instance of PostgreSQL. The entire app, binaries and all, lives in your Python program’s virtual environment. It does add about 30MB to the base footprint of the venv, but the resulting convenience is hard to match.
MongoDB
MongoDB binaries are available without an installer, so in theory they can be run from any directory. But if you want tighter integration with a given application without a formal install, that’s possible.
What the pgserver library does for PostgreSQL and Python, portable-mongodb does for MongoDB and Node.js. Add it to a Node project and you’ll have a private, automatically deployed instance of MongoDB for that project, or for other project environments if you so choose.
If you want to do this yourself, and want more granular control over the deployment, Atikur Rahman has published step-by-step instructions for creating a self-contained MongoDB instance. It’s less complex than you might think.
Redis
On Linux, Windows, and macOS, a common way to run Redis is to simply use a Docker container, or to install by way of a package manager (APT, RPM, Snap, Homebrew). But for a fully portable experience without Docker on Windows, a third-party build of Redis provides standalone Windows binaries. These are built directly from source and kept in synchrony with recent releases.