Missing features to support OData as an interface for Graph databases

[GEricson]

Using this CoPilot prompt with GPT5:
I am a system management software implementer. Create a comprehensive list all interfaces to graph databases as an Excel worksheet with name, sponsor, description, open source, license type, ISO/W3C standards, describe the reasons why OData v4.1 with the Aggregation and Temporal extensions cannot support this interface, then correlate the reasons into another Excel worksheet with columns for the reason, a detailed description of that reason, and a description of mitigation strategies for each reason. Finally export both worksheets as an Excel file.

graph_db_interfaces_vs_odata_updated.xlsx

[mikepizzo]

OData is focused on the data, not on the edges.

OData works extremely well for providing structured data to AI through things like MCP.

OData does not currently focus on the edges -- querying the edges, distances between nodes, etc.

Early versions of OData did explicitly model relationships, but we simplified in OData V4 to simple navigation properties.

You can model relationships as entities in the model, giving you the ability to add properties to those entities.

Microsoft Azure Search positions itself as a good option for RAG; it uses the OData protocol. We can look to how it models the data if we want to better understand how OData may be used in practice for RAG.