At its core, Rerun is a database. The OSS server is a local catalog that keeps catalog metadata in memory and reads data from RRDs with a footer on demand. It can therefore serve datasets larger than RAM, while Rerun Hub provides managed persistent storage for production use.
In this three-part guide, we explore a query workflow by implementing an "open jaw detector" on top of our face tracking example. This process is split into three steps:
- Explore a recording with the dataframe view
- Export the dataframe
- Analyze the data and send back the results
If you just want to see the final result, jump to the complete script at the end of the third section.