Duckservability is a DuckDB database schema for querying standard Linux systems via the /proc virtual file system which exposes real time kernel and process statistics. Traditionally files exposing these statistics have been aggregated via unix commandline tools. This project wraps these files in database tables that can be queried, joined, etc via SQL.
Duckservability is primarily meant as a proof of concept. The overarching goal is to illustrate a class of things that can be done when the traditionally coupled storage and compute components of a database system are decoupled.
Note that Duckservability presently only works on Linux and other Linux like systems.
- Install DuckDB
- Download the latest release of the DuckDB file containing the Duckservibility schema.
- Start DuckDB by running
duckdb duckservability.duckdb - Start querying! To do so, run the
show tables;command, or run one of the example queries.
Below are a few examples of questions you can ask with Ducservability. Additionally, see the example-queries directory for more examples.
Query processes that have performed the most IO operation in megabytes:
Duckservability is presently in a proof of concept state. If you're looking for a more mature SQL system for querying OS data, take a look at the osquery project.