Native Projections now in Aura Graph Analytics

We are excited to announce support for

Native Projections

in Aura Graph Analytics (AGA)! You can now efficiently project graphs directly from your underlying database storage into your GDS session using either standard Cypher commands or the GDS Python client.

What's new?

Previously, loading graphs required query-based approaches that, while flexible, weren't always optimized for sheer volume. With native projections, you can explicitly specify node labels, relationship types, and their associated properties to execute high-performance bulk graph loading.

Why this matters ⚡

Because native projections read directly from the underlying database storage rather than executing complex Cypher filtering, they offer

the absolute best performance and scalability for bulk graph loading

. This is the ideal tool when you need to project entire databases or massive subgraphs into memory quickly, without needing complex topological transformations.

How to use it 🛠️

Native projections are fully integrated into both your standard Cypher workflow and the GDS Python client (v1.22+). You can start using it right away:

Via Cypher:

CALL gds.graph.project(
  'my_native_graph',
  'Person',
  'KNOWS',
  {
    memory: '2GB'
  }
)

Via Python Client:

G, result = gds.v2.graph.project_native(
    graph_name="my_native_graph",
    node_label_filter=["Person"],
    relationship_type_filter=["KNOWS"]
)

For complete details on configuring these projections, supported parameters, and underlying mechanics, please check out the official documentation on Native projections in Aura Graph Analytics.