A manifest contains the location, data type, and details of user selection. Manifest locations are envisioned to be used primarily for identifying data in object-storage systems, either as direct object access or to data containers (zarr, tiledb, iceberg, etc.) at object-store prefixes.
A web service returning a manifest allows the consumer to access the data according to their use case:
- directly from high performance storage,
- in parallel,
- in the order desired,
- only a subset of the initially selected data
This pattern is particularly useful for large data volumes that are impractical to return in a synchronous access request. It is also very useful for large scale parallel processing.
may2025_ccp_fdsnws_example.jsonMay 2025 dataselect-api WIP exampletiledb_content_stub.jsontileDB stub