> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.nvidia.com/aerial/aodt/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.nvidia.com/aerial/aodt/_mcp/server.

# Results and Data

AODT simulation outputs are written as structured data products that can be
queried or inspected after a run. Batched simulation results are stored as S3
Parquet tables and can be queried from the catalog. RAN waveform dumps are
optional H5 files for inspecting received I/Q samples.

Use this section to understand what files and tables are produced, where to find
them, and how to interpret their fields.

## Storage and Catalog

Exported simulation outputs are written as Parquet files in S3-compatible
storage and registered in an Iceberg catalog. The S3-compatible store contains
the data files, while the catalog stores table metadata, including table names,
schemas, snapshots, and the locations of the Parquet files.

In the default local deployment, MinIO provides S3-compatible storage and Nessie
provides the Iceberg REST catalog at `http://nessie:19120/iceberg`. In AWS
deployments, Amazon S3 provides storage and AWS Glue provides the catalog.

![Storage and catalog flow diagram](https://files.buildwithfern.com/aerial.docs.buildwithfern.com/aerial/aodt/e6100e4a7f74e9fd2f7026f5a18a7a29cd333bbb927f088550b665bde4033f40/docs/assets/storage-catalog-flow.png)

The scripts in `client/examples/example_query_tables.py` and
`client/examples/example_drop_database.py` show example code for querying
the catalog and tables, and how to drop test data when it is no longer needed. You can adapt the table-querying code for your
own analysis or post-processing workflows.

Reference the exported Parquet tables, including setup records, channel
results, ray paths, telemetry, and RAN configuration fields.

Enable RAN time-frequency grid dumps and read the generated H5 datasets with
`h5py`.

## Related Workflows

* Use [Batched Mode](/batched-mode) to produce complete catalog-queryable result tables for all time steps.
* Use [Interactive Mode](/dynamic-mode) when your application needs selected results during a run.