FalkorDB V4.8: Neo4j requires 7x the memory to hold the same dataset

The latest FalkorDB release focuses on optimizing resource utilization, reducing memory footprint, and accelerating query execution.

With a memory efficiency that outstrips Neo4j—requiring 7x less memory to manage the same dataset—version 4.8 empowers users to handle complex graph workloads with unprecedented scalability and cost-effectiveness.

Key highlights of the release include:

• 42% Memory Reduction: Enhanced memory management allows for larger graph deployments on compact hardware, broadening accessibility for enterprises and developers.
• 65% Faster Aggregations: Optimized functions like COLLECT cut processing times significantly. In benchmark tests with the query UNWIND range(0, 1000000) AS X RETURN x % 1024, collect({x:x, minus_x: -x}), v4.8 reduced execution time by 65%.
• Full-Text Edge Indexing: New support for full-text indexes on edges enhances search functionality and enables more sophisticated graph traversals based on relationship properties.
• GraphBLAS Upgrade: Integration of a new GraphBLAS version with 32-bit matrix indices support.

Check it out (Start free): FalkorDB Graph Database

Comments

[Dan Shalev]

I'd add that Aggregations in general have been improved. For example, the COLLECT function has been optimized to reduce the time it takes to collect items in a list. For example, in tests using the query UNWIND range(0, 1000000) AS X RETURN x % 1024, collect({x:x, minus_x: -x}), version 4.8.3 showed a 65% reduction in processing time.