I work on Postdot, an open-source, self-hosted social media scheduler. Before choosing where to run one, I wanted a number: how much memory does a scheduler hold while nothing is happening? I measured two, with three fresh installations each, on the same machine.
What I measured
- Postdot 0.1.0-alpha.7: one container, SQLite and local files.
- Postiz v2.25.3, started from its official Docker Compose file: eight containers. The application, PostgreSQL, Redis, and a Temporal stack with its own PostgreSQL, Elasticsearch, web interface and admin tools.
Both used images pinned by digest, started from empty volumes, with no account, channel or post. After a settling period (240 seconds for Postiz, 60 for Postdot) I took 16 docker stats samples per container, 15 seconds apart. The host was a Ryzen 7 5800X with 32 GB and Docker 29.9.3, with no other containers.
Results
| Idle memory (mean of three runs) | |
|---|---|
| Postdot, one container | 83 MiB |
| Postiz, application container | 1,303 MiB |
| Postiz, complete Compose stack | 2,303 MiB |
Postiz's own documentation lists 2 GB of RAM as the supported floor for its all-in-one Compose stack, so the 2.3 GiB I measured is consistent with it. The runs agreed closely: Postdot between 82.9 and 83.9 MiB, the Postiz stack between 2,296 and 2,308 MiB.
How to read it
If you already run PostgreSQL, Redis and Temporal, Postiz documents connecting to them instead of deploying its bundled copies. In that case the fair comparison is application against application: 1,303 MiB against 83 MiB. If you start from nothing, it is the complete stack.
This measures idle memory only. It says nothing about load, long uptime, many channels or large media, and the two products differ in features and maturity: Postdot is in alpha.
Full method
The article on the Postdot blog has the complete method, the image digests, the scripts and the raw samples: https://postdot.app/blog/postdot-postiz-idle-memory?utm_source=devto&utm_medium=social&utm_campaign=idle-memory
To try Postdot: https://postdot.app/docs/installation/?utm_source=devto&utm_medium=social&utm_campaign=idle-memory
Drafted with AI assistance, then checked and edited by me against the measurements.